5 September 20091

1Au, W. (1989) Target detection in noise by echolocating dolphins In Thomas, J. & Kastelein, R. eds.Sensory Abilities of Cetaceans. NY: Plenum Press pp 203-216

Dolphin Biosonar EcholocationA Case Study

byJames T. Fulton

..(Appendix L on the website)

Structure without function is a corpse; function without structure is a ghost

Vogel & Wainwright, ‘69

1.0 A case study in echolocation and communications in the dolphin

A Table of Contents and an Index are available at the end of this Appendix.

The order Cetacea (whales, porpoises and dolphins) is well known for its broad use of its uniqueunderwater capability in active echolocation and communications. The capability appears to beparticularly well developed in the dolphin, suborder Odontoceti, and particularly in the familyTursiops of dolphins. This study will focus on the most carefully studied member of this family, thewell-known bottlenose dolphin, Tursiops truncatus.

Au has noted1, “The use of acoustic energy is probably the most effective way to probe an underwaterenvironment for purposes of navigation, obstacle avoidance, prey and predator detection and objectlocalization and detection.” It also appears optimum for intraspecies communications andintraspecies peer tracking.

Much of the material on the neural system in this study is drawn from a much more comprehensivetreatise on hearing available on the Internet, “Processes in Biological Hearing,” atwww.hearingresearch.net. That material is voluminous. A guide to that material is to be publishedduring 2007 as “Biological Hearing: A 21st Century Tutorial.” The above website will highlight thepublication and availability of the guide. Paragraph numbers shown below in square brackets referto paragraphs in the downloadable chapters on the website (ca. 2006).

1.1 Overview of hearing and active echolocation in Dolphins

Dolphins rely upon their active and passive sonar capabilities for their livelihood. Consequently, theperformance of these systems is highly developed, and probably near optimum in performance.The acoustic capabilities of the dolphins have been studied extensively (partly because of theirmilitary potential and their ability to influence military hardware design). However, Cranford madean important statement in 2000. “Our current understanding of aquatic echolocation by odontocetes,

5 September 20092

2Morris, R. (1986) The acoustic faculty of dolphins In Bryden, M. & Harrison, R. eds. Research onDolphins. Oxford: Clarendon Press chap 18

especially in free-ranging animals, is in its infancy.”

A review of the unclassified literature suggests the field has not employed a system analyst insupport of the extensive past empirical programs. The current literature is polarized as to the overallacoustic capability of the dolphins. One camp supports a capability with the larynx as the acousticsource. Another supports a capability that deprecates the larynx as a source. Morris has provided anoverview of this controversy as of 19862.

A system analyst, without an in-depth review, would immediately suggest that the dolphin probablyuses both capabilities. It would use the lower frequency whistles for long range, and possibly Dopplerdiscrimination, due to the lower attenuation of water in that frequency region. Alternately, it woulduse higher frequency clicks for accuracy, particularly in angular spatial resolution, where range wasnot an issue. It would seem unusual for the dolphin to give up the low frequency echo detectioncapability that it brought with it from its terrestrial days. Even humans can make crude rangeestimates by yelling in a rock canyon and estimating the time the echo is heard.

From a system design perspective, the biosonar of the bottlenose dolphin, Triosops truncatus, isprobably the most sophisticated target location and analysis system in existence. It at least equalsthe performance of the most advanced military sonars and radars. The sonar of the bottlenosedolphin can be described as a multifunction, multifrequency, multimode, physically adaptablemonopulse sonar with extensive pattern matching capability and optimal interference rejection. Inaddition, it offers both binaural receiving and biphonic (or bivocal) transmitting capabilities. As aresult, the dolphin has a short range echolocation and analysis capability at less than 100 metersthat is not matched by any man-made system. Within the range of 100 to 600 meters, the dolphin hasa highly optimized system tailored to its particular ecological niche, and not necessarily comparableto any man-made system. The dolphin apparently has no need for echolocation beyond 600 meters. Itmay have a need for communications (and potentially direction finding) at ranges greater than 600meters. The features described above for the bottlenose dolphin are packaged within a volume of one cubicfoot and offers all of the features of the Ku band radar on the Shuttle space transportation system. As in the case of the Shuttle, the dolphin echolocation system can also be used for intra-speciescommunications. The system is used for this function so extensively, and their repertoire of signals isso wide, the dolphins are sometimes described as "sea canaries" like their better known cousins theBeluga whales. This analysis explores the anatomy, morphology, physiology and performance of the system of thebottlenose dolphin in considerable detail.

The controversy over the larynx versus the upper nasal passages as a high frequency acoustic source(frequencies greater than 30 kHz) was not resolved in a comprehensive analytical manner until theintroduction of high speed X-ray and MRI techniques in the 1990's. The resolution was dramatic andshowed that most of the descriptive material in the Cetacea literature prior to 1990 must bediscounted.

5 September 20093

3Au, W. (2004) The dolphin sonar: excellent capabilities in spite of some mediocre properties In Porter, M.Siderius, M. & Kuperman, W. eds. High Frequency Ocean Acoustics. Melville, NY: AIP ConferenceProceedings

4Abawi, A. Hursky, P. Porter, M. et al (2004) Biomimetic signal processing using the biosonarmeasurement tool (BMT) In Porter, M. Siderius, M. & Kuperman, W. eds. High Frequency OceanAcoustics. Melville, NY: AIP Conference Proceedings pp 260-271

5Norris, K. Wursig, B. Wells R. & Wursig, M. (1994) The Hawaiian Spinner Dolphin, Berkeley, Ca:University of California Press pg 169

Recent statements by Au3, quoted in Abawi, et. al., appear to reflect the lack of an overall systemanalysis of the acoustic capabilities of the dolphin. “Despite their impressive performance, thedolphin’s reception and transmission subsystems are quite mediocre compared to its signalprocessing capabilities4.” On its face, this statement appears to reinforce the timeliness of ProfessorNorris’s position; “One fact has consistently emerged from recent biological studies, particularly inthe spheres of biophysics and anatomy, and it is that we humans have, in the past, usuallyunderestimated the refinements of animal adaptation. Accordingly, if we are to establish a workinghypothesis for cetacean echolocation and its anatomical correlates, it is probably better to expectsophistication . . . “ (Norris 1964, p. 324).

From an analyst’s perspective, the literature suggests a much more capable reception andtransmission system than many perceive. It is also important to appreciate that the signalprocessing system can only process information received from the reception and transmissionsystems. The sonar of the bottlenose dolphin is considerably more sophisticated than any currentman-made sonar in the world. It rivals the most advanced airborne radars available today. Merelylisting the most obvious properties of this biological sonar (biosonar) attests to its fantasticcapabilities. It is fundamentally a multi-band, multi-mode (including Doppler detection) frequency-hopping, steerable beam, binaural receiver, camouflage penetrating, single-pulse (when required)system with properties at least as sophisticated as the latest stealth fighter plane, the F-117, andlatest stealth bomber, the B-2. The system is also multi-application, serving as both a sonar systemand an intra-species communications system, like the Ku-band radar of the Space Shuttle.

This paper will surface a variety of system features that support the above description of thecombined transmission, reception and data processing systems of the bottlenose dolphin inparticular. Many of these features were only suggested obliquely in the literature prior to the year2000.

A major problem in discussing the functionality and capability of the dolphin’s hearing is theextremely small number of animals that have been studied. In many cases, sweeping conclusionshave been drawn by investigators working with a single animal. Averaging of data, related to anyparameter, from more than five specimens is a rarity. As shown in some of the graphs of Norris, etal, the variation in a parameter can approach twice the mean5. In such cases, the data is asrepresentative of a dichotomy as it is representative of a Gaussian variable. In the case cited, thetypes of sounds produced by eight specimens were examined.

Because of the few specimens examined by any individual investigator, the description of themorphology, physiology and performance of the species presented here will necessarily be tentative.

A major point regarding echolocation in marine mammals relates to the method of range detection. Man-made sonar machines rely heavily upon measuring the time of travel of a pulse to determine the

5 September 20094

6Reynolds III, J. Wells, R. & Eide, S. (2000) The Bottlenose Dolphin. Gainesville, FL: University Press ofFlorida

7Ridgway, S. (1995) Dolphin Doctor, 2nd Ed. San Diego, Calif: Dolphin Science Press pg 81

8Nachtigall, P. (1986) Vision, audition, and chemoreception in dolphins and other marine mammals InSchusterman, R. Thomas, J. & Wood, F. eds. Dolphin Cognition and Behavior: A Comparative Approach. Hillsdale, NJ: Lawrence Erlbaum Associates Chapter 4

9Supin, A. Popov, V. & Mass, A. (2001) The Sensory Physiology of Aquatic Mammals. Boston, MA:Kluwer Academic Publishers Chapter 4

10Ridgway, S. & Harrison, R. (1981) Handbook of Marine Mammals, Vols 5 & 6. NY: Academic Press

absolute range to a target. The times to be measured are on the order of microseconds and requirecircuit bandwidths greater than 100,000 Hz. High stability oscillators (at least a few parts in 108),and complex counters are de rigueur, and an absolute range is required so its value can becommunicated to some other human at a distance. Known mammalian biology does not supportcircuit bandwidths greater than 1000 Hz. Because of this great discrepancy, the investigator shouldfocus on other methods of range determination in Cetacea. A method that overcomes the neurologicalbandwidth limit is the use of phase locking in a “servomechanism loop.” The oscillation used as areference in this servo loop provides a relative indication of distance to the target without using anyhigh frequency oscillators, counters or other circuits of Man’s technical toolbox. A relative range isall that is needed by an autonomous biological system.

Based on the above brief discussion, the following should be noted. Cetaceans do not use precisiontime measurement techniques in ranging and, like other mammals, they do not use transcendentalcalculations in angle measurements.

1.1.1 Species specific background

As a brief aside, the vision of the dolphin has not been investigated at a detailed level and theliterature contains a variety of anecdotal comments. There are frequent assertions that the dolphinis nearsighted out of water, in analogy to the human being farsighted in water (Reynolds, p. 81)6. The analogy is questionable because the dolphin does not depend on its cornea for a major part of theoptical power of its visual imaging system. The outer cornea is nearly flat and of constant thicknessexcept at the very edges. Caldwell & Caldwell (pg 107) insist on the excellent vision of the dolphinout of water. They illustrate this by showing a dolphin leaping more than 6 meters into the air tosnatch a small fish from the teeth of a performer at an aquarium. A scenario in Ridgway suggests adolphin may recognize a specific person at about 20 meters using either vision or hearing in air7. It isnoted that the two eyes of the dolphin, which are located on opposite sides of the head, do rotateindependently. The dolphin appears to have some forward (binocular) vision but it appears to belimited. It may or may not involve a fovea as in many other mammalian species. Nachtigall, et. al.provided the best description of dolphin vision up to 19868. The inconsistency in the available data isobvious in that paper. Supin, et. al. provided a much better description in 20019. It showsspecifically how the bottlenose dolphin eye has evolved differently than the human eye and isuniquely designed to maintain proper focus whether in or out of the water.

1.1.1.1 Literature

Ridgway & Harrison have provided a two volume set of data concerning dolphins in handbook form10.

5 September 20095

11Barnes, L. (1990) Fossil record and evolutionary relationships In Leatherwood, S. & Reeves, R. eds. TheBottlenose Dolphin. NY: Academic Press pg 10

12Heyning, J. & Mead, J. (1989) Evolution of the nasal anatomy of Cetaceans In Thomas, J. & Kastelein, R.eds. Sensory Abilities of Cetaceans. NY: Plenum Press pp 67-79

13Supin, A. Popov, V. & Mass, A. (2001) The Sensory Physiology of Aquatic Mammals. Boston, MA:Kluwer Academic Publishers pg 25

14Morris, R. (1986) The acoustic faculty of dolphins In Bryden, M. & Harrison, R. eds. Research onDolphins. Oxford: Clarendon Press Chap 18

Barnes, in Chapter 1 of Leatherwood & Reeves, reviews the phylogeny of order Cetacea (whales,dolphins, and porpoises) and the evolutionary sequence of changes related to the dolphins of thefamily Tursiops in particular11. It appears these mammals left land between 70 and 90 million yearsago. The dolphins are members of the suborder Odontoceti (whales that possess at least one row ofteeth as adults). There are four super families within Odontoceti; including the largest,Delphinoidea. Delphinoidea includes the Tursiops family. Cetacea are most closely related to thehippopotamus and other ruminants having multiple stomach chambers. The dolphins swallow theirfood whole. Heyning & Mead have provided additional details relative to the evolution of the nasalstructures of Cetacea12.

Ridgway, in Chapter 4 of Leatherwood & Reeves, reviews the morphological features of the brain andcentral nervous system of the bottlenose dolphin, Tursiops truncatus. Truncatus refers to thesignificant wearing of the teeth in adulthood. He makes a number of comparisons with the brain ofhumans, which it is of comparable brain weight to total weight. He notes that, contrary to thehuman literature, the brain of the dolphin is clearly more convoluted than that of humans. Withregard to the auditory system, Ridgway references Bullock & Gurevich. They note the size of themedial geniculate is about seven times that in humans, the inferior colliculus is about 12 times itshuman analog, and the nucleus of the lateral lemniscus is more than 250 times as large as theequivalent structures in humans. The ventral cochlear nucleus is also massive compared to thehuman analog. The auditory nerves are similarly larger than in comparable humans (70,000 to100,000 neurons with mean diameters of 12 microns13). The visual system, on the other hand is lesswell developed in these animals with no obvious fovea centralis. Because of packaging changes, thelocation of activity on the cerebral cortex appears significantly different from in humans. The analogof layer four (the granular layer) of the human cortex is absent or difficult to locate in the dolphinbrain.

The detailed character of the sound generation and sound receiving systems of the dolphin, andCetacea in general, remain poorly documented. Morris provides a comprehensive listing of thefeatures of these systems (in 1986) interpreted by two different schools (the American and theEuropean, p. 380-381) and resulting in two widely differing opinions of how the echolocation systemin dolphins operates14. The interpretations were far from compatible in 1986. He addresses thisdichotomy finally on page 393. Morris closed with a comment concerning the slowness of futureprogress in observing the dolphin in their natural habitat unless some new and unexpectedtechniques should arise. The recent development and use of self-contained, GPS-based, navigationalrecording and reporting capabilities along with new miniaturized acoustic data recording capabilitiesappear to have solved this problem.

The literature notes the asymmetries in both the overall configuration of the forward torso of thedolphin and specifically in the melon within the forehead area (Au, p. 89). This appears to be similar

5 September 20096

15Cranford, T. Amundin, M. & Norris, K. (1996) Functional morphology and homology in the odontocetenasal complex: Implications for sound generation J Morphol vol 228, pp 223-285

16Cranford, T. (2000) In search of impulse sound sources in Odontocetes In Au, W. Popper, A. & Fay, R.eds. Hearing by Whales and Dolphins. NY: Springer Chapter 3, pg 134

17Houser, D. Finneran, J. et al. (2004) Structural and functional imaging of bottlenose dolphin (Tursiopstruncatus) cranial anatomy J Exp Biol vol 207, pp 3657-3665

18Caldwell, M. Caldwell, D. & Tyack, P. (1990) Review of the signature-whistle hypothesis for the Atlanticbottlenose dolphin In Leatherwood, S. & Reeves, R. eds. The Bottlenose Dolphin NY: Academic Press Chap 10

19Ridgeway, S. (1995) Dolphin Doctor, 2nd Ed. San Diego, CA: Dolphin Science Press pg 51

to the asymmetry found in the outer ears of owls, and possibly bats. However, in the case of thedolphin the asymmetry is in the transmitting mechanism instead of the receiving mechanism. Inboth cases, the asymmetry probably introduces a vernier into the spatial location determination. Aslight vertical motion of the head introduces a slight horizontal change in the predicted targetlocation. Cranford, Amundin & Norris provide the broadest possible analysis of the anatomy of theOdontocete related to sound generation. They used a variety of X-ray tomography, MRI, and varioushistological techniques to study 40 specimens from 19 species15. An asymmetry in the ultra highfrequency energy generating system is also noted. The material is extremely useful but theirproposals relating to the mechanism of sound generation are only first-order. Cranford has describedthe right-hand energy generator, within the right nare, as being twice the size of the left-handgenerator16.

The asymmetry in the dolphin head is much smaller than the amount of space given to the subject inthe literature. Houser, et. al. have provided imagery and a discussion supporting this point17. Theasymmetry is perceptible but not obvious to the non-anatomist.

Caldwell, et al. have provided a good review of the low frequency signaling by the bottlenosedolphin18. The purpose of these whistles has not been adequately described. Caldwell et al. describeddistress and signature whistles that suggest communications. She also describes concern whistlesassociated with a change in depth of the water in its enclosure. While these whistles can beconsidered distress calls, they can also be considered investigative whistles, re-measuring theavailable size of the enclosure. Ridgway has documented even clearer examples of dolphincommunications, both through water and through air19. The vision of the dolphin is comparable tothat of the human (recognizing an individual person from several hundred feet away).

The Ridgway book was prepared for the popular press. However, it contains many interestingphysiological facts about the bottlenose dolphin. These facts include:< comments on communications and sound imaging on pages 56-61 & 87,< maximum diving depths exceeding 1000 ft on page 11,< maximum speeds on the order of 25 miles per hour (40 kilometers per hour) on page 70,< maximum heart rates of 100 beats per minute after deep diving or maximum exertion,< resting heart rates of 35 beats per minute on page 97,< typical respiration interval of 30 seconds but it can hold its breath for up to four minutes,< oxygen level in lungs reduced to 2% after three minutes and,< excellent vision in air (identifying humans at several hundred feet distance) on page 80.

5 September 20097

20Green, R. Ridgway, S. & Evans, W. (1980) Functional descriptive anatomy of the bottlenose dolphinnasolaryngeal system In Busnel, R-G. & Fish, J. eds. Animal Sonar Systems NY: Plenum Publishing pgs199-238

21Ridgway, S. Carder, D. Green, R. et. al. (1980) Electromyographic and pressure events in thenasolaryngeal system of dolphins during sound production In Busnel, R-G. & Fish, J. eds. Animal SonarSystems NY: Plenum Publishing pgs 239-249

22Au, W. (1993) The Sonar of Dolphins. NY: Springer-Verlag

23Cranford, T. Amundin, M. & Norris, K. (1996) Functional morphology and homology in the odontocetenasal complex: implications for sound generation J Morph vol 228, pp 223-285

24Au, W. (2004) The dolphin sonar: excellent capabilities in spite of some mediocre properties In Porter, M.Siderius, M. & Kuperman, W. eds. High Frequency Ocean Acoustics. Melville, NY: Amer Inst Physics

25Floyd, R. (1986) Biosonar signal processing applications In Nachtigall, P. & Moore, P. eds. AnimalSonar: Processes and Performances. NY: Plenum Press pg 774

A team including Ridgway produced two important papers in 1980 on the musculature in the head ofthe dolphin and the pressures within the passageways during phonation20,21. The data describing thesounds produced during testing is less complete.

Au has presented a wealth of information on the high frequency capability of the combined aural andhearing systems of the dolphin Tursiops truncatus, and other odontocetes22. The high frequencysignals are dominated by short clicks.

Currently, the definitive description of the morphology of the bottlenose dolphin is that of Cranford,Amundin & Norris of 199623. It is based on in-vivo CT and MRI scanning. The report is extensive,figures are well annotated, supported by an extensive bibliography, provides tabular values for manyelements and contains many figures in color. The goal of the study was to provide quantitative dataon the size, shape, degree of asymmetry and geometric configuration for the components of theimpulse sound generation mechanism in the odontocete forehead. They provide a concise discussionof the terminology found in the literature. Although the authors indicate they collected data on alarge number of specimens (~40 representing 19 species), the number for any one species is quitesmall. In addition, they note that they do not have both morphology data and sound recordings fromthe same animal.

Unfortunately, when they begin to discuss the physiology of the various members of Cetacea based ontheir data, they limit the discussion to click generation (as opposed to continuous tone and otherpotentially communications-oriented sounds).

Au presented an unusual paper in 200424. It described the signal generation and signal receivingcapabilities of the dolphin as mediocre on three occasions without providing a detailed description ofthe actual capability of the species. This work suggests that the capabilities are, on the contrary,quite sophisticated. Au made two observations in his discussion section. “However, within a 100 mrange, there is not a technological sonar that can rival the dolphin in discriminating and recognizingtargets.” He continued later. “. . .and, we should strive to produce a short-range sonar system thatcan perform as well as the dolphin.” This work suggests much of that capability is due to thesophisticated signal generation and signal receiving capabilities of the dolphin, in addition to itssignal processing and cognitive abilities. Floyd, a colleague of Au, has shown that the figure of meritfrequently used to evaluate man-made sonars does not give a rational value for the dolphin25. A

5 September 20098

26Lemonds, D. Au, W. Nachtigall, P. Roitblat, H. & Vlachos, S. (2000) High frequency auditory filtershapes in an Atlantic bottlenose dolphin J Acoust Soc Am vol 108, pg 2614(A)

27Cranford, T. (2000) In search of impulse sound sources in Odontocetes In Au, W. Popper, A. & Fay, R.eds. Hearing by Whales and Dolphins. NY: Springer Chapter 3

careful reading of Floyd’s paper shows his performance estimates are not based on the physiology ofthe dolphin, particularly its associative correlation capabilities, but on estimates based on simpletechniques used in common man-made sonar systems (See also Section 1.5). He does note, thewideband echolocation signal (apparently pulse mode) and binaural receivers “provides for superiortarget recognition capabilities, better target detection in reverberation, and far better spatialresolution than conventional sonars with similar beamwidths”

Au’s computation of the equivalency between a long constant amplitude tone burst and the shapedtone burst of the pulse-based dolphin system appears irrelevant. While the energy in the longerburst is obviously higher, and its detectability at longer ranges would be higher, other performanceaspects of the dolphin system would be seriously compromised by the use of such a long burst.

Au provides some critical band measurements from Lemonds, et. al26. These measurements do notrepresent the minimum frequency difference sensitivity of the dolphin. It can be shown theyrepresent the equivalent noise bandwidth of the multidimensional correlator of the dolphin PGN(Section 1.4.4). The minimum detectable frequency difference is much narrower than the criticalband value (Section 1.4.2.4.2).

Cranford has recently provided a chronologic look at the quest to understand the biosonar system ofmarine mammals27.

1.1.1.2 Brief Glossary

As Cranford, Amundin & Norris note on page 224, “In the fusiform cetacean body architecture,posterior is synonymous with caudal, anterior is synonymous with rostral, dorsal is synonymous withsuperior, and ventral is synonymous with inferior.” They also note that many terms have entered thelexicon of Cetacea that were not developed by anatomists. A brief list of terms important to thisstudy is summarized below. A few of these terms have meanings unique to this study.

Ablaut– The Proto-Indo-European system of root vowel alternations.

Allophone– The different sounds that can represent one phoneme in the speech of a given speaker orlanguage; that is, they are perceived under certain circumstances to be the same phoneme.Allophonic systems can vary from speaker to speaker or more especially from language to language.e.g., /s/ can be represented by the allophones /s/ and /z/ (sound or sounds).

Articulation– The point of production of the various segments in the oral cavity.

Biosonar– A biologically based system of sound source location by determining its angle and range. The system may be passive or active.

Bound morpheme– A morpheme which never occurs alone, but is attached to other morphemes.e.g., Eng. kindness, unlikely, Ger. Mädchen, inflectional endings, etc.

5 September 20099

Bursae– A sac or sac like body cavity, especially one containing a viscous material. The materialmay constitute a liquid crystal and exhibit significant structure in its acoustic role.

Commissure– Used in two contexts in anatomy.1. A tract of nerve fibers passing from one side to the other of the spinal cord or brain. 2. The point or surface where two parts, such as the eyelids, lips, or cardiac valves, join orform a connection.

Consonant– Any segment produced by stopping and releasing the air stream (stops), or stopping itat one point while it escapes at another (liquids), or a very narrow passage causing friction(fricatives).

CT– Primarily X-ray based computed tomography. See tomography.

Dialect– 1. A regional variety of a language distinguished by pronunciation, grammar, or vocabulary,especially a variety of speech differing from the standard literary language or speech patternof the culture in which it exists. 2. A language considered as part of a larger family of languages or a linguistic branch:Spanish and French are Romance dialects.

Diphthong– Syllabics which show a marked glide from one vowel to another, usually a steady vowelplus a glide. e.g., /ou/ in house, /oi/ in toy.

Fricative– A consonant produced when the air released by an articulator passes through a narrowpassage with audible friction. e.g., [f], [s], [þ], [ð], etc.

Globular bushy cells– A morphological designation for the stage three signal encoding neurons inthe ventral cochlear nucleus (VCN). Used in Zook & DiCaprio, 1990.

Gular– The throat.

Junk– A description (from the whaling industry) of the massive fatty tissue within the forehead ofCetacea.

Language– 1.The use of acoustic sounds in organized combinations and patterns in order to expressand communicate thoughts and feelings.

2. An organized set of combinations and patterns of acoustic sounds used by a specificpopulation to communicate thoughts and feelings.

Lexicon– Linguistics. The morphemes of a language (or dialect) considered as a group.

Liquid crystal– A unique state of matter that exhibits zero long term resistance to an appliedpressure (will flow to fill a container) but can exhibit significant impedance to short term distortingforces. Such materials were frequently described as gels in the earlier literature.

Magnetic Resonance Imaging (MRI)– A method of stimulating individual molecules, causingthem to vibrate at their molecular frequency, and recording (imaging) their density at a given

5 September 200910

location based on that vibration.

Melon– A description (in the vernacular) of the massive fatty tissue within the forehead of Cetacea.

Monkey lips– (a.k.a. phonic lips) A description (in the vernacular) of the valvular flaps located ineach nasal passage within the head of Cetacea. Apparently capable of generating both fricativesounds and impulses by slapping the lips together.

Morpheme– The smallest unit of meaning. Any word or part of a word that conveys meaning andcannot be further divided into smaller meaningful elements.

Morphology– The study of forms of language, especially the different forms used in declensions,conjugations, and wordbuilding.

Phoneme– The simplest significant unit of sound. A phoneme may also have various allophones.

Phonetics– A branch of linguistics dealing with the analysis, description, and classification of speechsounds, or segments. Phonology– The branch of linguistics concerned with the structural relationships betweensegments. The study of phonetics and phonemics together in the evolution of speech sounds.

Pidgin– A term used by linguists to define the first few words developed by two people seeking toconverse but not speaking a common language.

Principal call A morphological designation for the stage three signal decoding neurons in themedial nucleus of the trapezoidal body (MNTB). Used in Zook & DiCaprio, 1990.

Sac– A pouch or pouch-like structure, sometimes filled with fluid or air. See theca.

Theca– A case, covering, or sheath holding something other than air. See Sac.

Tomography– A method of preparing detailed images of a predetermined plane section of a solidobject while blurring out the images of other planes.

Unvoiced– A segment produced without any accompanying vibration of the vocal cords. e.g., [p] v.[b], or [t] v. [d], etc.

Voiced– A segment produced with accompanying vibration of the vocal cords. e.g., [b] v. [p], or [d] v.[t], etc.

Vowel– A voiced segment characterized by generalized friction of the air passing in a continuousstream through the pharynx and opened mouth, with relatively no narrowing or other obstruction ofthe speech organs.

1.1.1.3 Technical environment

5 September 200911

28Tolstoy, I. & Clay, C. (1966) Ocean Acoustics. NY: McGraw-Hill

1.1.1.3.1 The acoustic environment of sea mammals

The acoustic sea environment is very complex and difficult to quantify. Both surface conditions andbottom conditions have a significant impact on sound reception within the sea environment. Withinthe volume of the sea, there are also many inhomogeneous features which can scatter sound. Simultaneously, the attenuation of sea water increases with frequency. As a result, most oceancommunications are short range. An exception may be the very low frequency signaling (atfrequencies below one kilohertz) between whales at considerable distance from each other. Thesignaling rate at such low frequencies is necessarily very low.

Tolstoy & Clay have described both the attenuation and noise environments in the sea28. Figure1.1.1-1 reproduces their figure 1.1 with extrapolations applicable to the dolphin. The major variationin attenuation, and the two-way attenuation inherent in echolocation, suggests the higher frequencyranges ( 80-150 kHz) are only available for short range echolocation (less than 100 meters). Theseranges would probably be used only for precision ranging during predation. The range from 30-80kHz offers greater range for the same peak source power at proportionately less resolution. Itremains to be documented as to whether this frequency range is or is not associated with the range capability of 600 meters described above. The lower frequencies (200-1000 Hz) appear better suitedfor long range communications (such as across a large estuary). The communications source needonly be strong enough to provide adequate signal strength at the receiver by overcoming the one-wayattenuation. As noted above, frequencies below 200 Hz are thought to be used by the large whales toachieve very long range communications.

This figure can be extended to even lower frequencies to show why the larger members of Cetaceaapparently use frequencies below 50 Hz to signal over intercontinental distances (although suchsignalling may also require the use of ducts within the ocean layers to avoid the uncontrollablespreading of the energy in undesired directions).

5 September 200912

Figure 1.1.1-1 High frequency sound attenuation in sea water. The solid curve is calculated for 20°C and S = 35parts per thousand. Shaded area indicates experimental verification. The dashed curve and acoustic ranges are extrapolations by Fulton. 1 neper = 8.686 dB. Modified from Tolstoy & Clay, 1966.

5 September 200913

Figure 1.4 of Tolstoy & Clay, on the noise environment, requires too much interpretation toreproduce here. However, Au has provided a graphic that is more direct. It clearly shows the noisespectrum as a function of acoustic frequency due to both animate and inanimate causes. Au notedthe extremely high noise in Kaneohe Bay is due to the high concentration of snapping shrimp.

1.1.1.3.2 Thefrequency range ofdolphin emissions

The research communityhas not yet settled on auniform description of thephonations of dolphins. Cranford has discussed thisdifficulty (p. 117). Thiswork will use the followingterms:

1. Low frequency (LF)–Signals generated in thepneumatic arena (larynx)and audibly detectable byhumans withoutinstrumentation. Typicallyin the frequency range of200–15,000 Hz.

2. High frequency (HF)–Signals generated in the pneumatic or non-pneumatic arena (tissue and bones) by the larynx andpropagated primarily in the aquatic environment. Typically in the 4,000–30,000 Hz frequency range.

3. Ultra high frequency (UHF)– Signals generated primarily in the non-pneumatic arena by thephonic lips (valvular flaps) and propagated exclusively in the aquatic arena. Typically in the 30-150kHz frequency range.

This categorization separates the largely incidental pneumatically propagated LF sounds of thedolphin from its more significant HF and UHF sounds associated with echolocation. However, morerecent investigations with more modern equipment suggests the HF region may extend up to at least96 kHz.

The signals generated by the dolphin lips in the UHF are generally sinusoidal as expected by theringing caused by the clapping together of two hard objects. The HF and LF spectrums show muchbroader ranges of frequencies simultaneously (that may or may not be harmonically related).

1.1.1.3.3 The dispersion of water-based liquids

Figure 1.1.1-2 Ambient noise in the ocean measured in 1/3 octave bands. Deepwater noise for different sea states are shown for comparison. From Au, 1989.

5 September 200914

29Herzfeld, K. & Litovitz, T. (1959) Absorption and Dispersion of Ultrasonic Waves. NY: Academic Press

30Porter, M. Siderius, M. & Kuperman, W. ed. (2004) High Frequency Ocean Acoustics. NY: AmericanInstitute of Physics. Chapter 1.

31Morris, R. (1986) The acoustic faculty of dolphins In Bryden, M. & Harrison, R. eds. Research onDolphins. Oxford: Clarendon Press pg 388

32Au, W. (1989) Target detection in noise by echolocating dolphins In Thomas, J. & Kastelein, R. eds.Sensory Abilities of Cetaceans. NY: Plenum Press Pg 205

The dispersion observed in sea water has had a checkered history. Herzfeld & Litovitz wrote adefinitive work in the subject in 1959, even discussing the potential for water to form molecules morecomplex than H2O

29. While significant dispersion was reported in a number of papers immediatelyafter World War II, Herzfeld & Litovitz disparaged these reports (page 354). They make a strongstatement concerning pure water, i. e., water absent any dispersed solids (p. 428). “No velocitydispersion has been found.” Herzfeld & Litovitz also described their experimental method ofmeasuring the elusive (if present) parameter, dispersion, in a fluid (p. 365).

A review of the subsequent literature foundvery few comments concerning dispersion in seawater or fresh water. One exception is byBuckingham30 who was studying sedimentsaturated water. Such a material is clearly anon-Newtonian fluid (Section 4.1.4 in“Processes in Biological Hearing”).

The literature suggests that dispersion is not acharacteristic of concern in the propagation ofsound in sea water, at least up to frequencies of150 kHz.

1.1.1.3.4 Range and Rangeresolution in sea water

Figure 1.1.1-3 shows the round trip travel timefor an acoustic pulse in sea water and the timedifference resolution required to achieve a givenrange resolution. The maximum range of thebottlenose dolphin has been estimated from themaximum pulse interval commonly measuredfor its high frequency echolocation system. Morris reported this value as 0.6-0.8 seconds31. The maximum range of the ultra highresolution pulse echolocation system is on theorder of 100 meters for a 7.62 cm diametermetallic test sphere32.

The range resolution of the ultra high frequencyecholocation system of the bottlenose hasremained controversial. Values as low as 1.5-4

Figure 1.1.1-3 Range equation and range resolution in seawater.

5 September 200915

33Au, W. (1993) Op. Cit. pg 240

mm (requiring a time difference resolution of 1-3 μsec) have been reported. Such a value wouldrequire much broader neural signaling circuits than normally found in mammals or exceptionallysophisticated signal manipulation within the neural system.

Au confirmed the range resolution of the bottlenose was unknown in 1993. When speaking of theultra high frequency echolocation system, he suggested it was probably 12-15 μsec (1.0 to 1.1 cm)33.

1.1.1.3.5 Doppler frequency versus velocity in sea water

It is likely that the Doppler frequency resulting from the motion of targets illuminated by theecholocation capability of dolphins plays an important role in their overall acoustic performance.Figure 1.1.1-4 provides the relationship between frequency shift and relative target velocity in seawater. A nominal maximum velocity for the dolphin is shown along with a preliminary estimate ofthe minimum velocity difference required by the echolocation system of a dolphin to detect a movingtarget. These estimates will be justified in the following discussion.

1.1.1.4 Echolocation in Humans

Recently, discussions have appeared concerningthe ability of humans to use short rangeecholocation effectively. The Learning Channelrecently presented a program, entitled “The BoyWho Sees through Sound,” concerning BenUnderwood, a teenager blind since the age ofthree due to retinal cancers in both eyes. Beginning at that age, he began experimentingwith clicks generated between his tongue andteeth with the goal of understanding hissurrounding environment. Using this largelyunobtrusive source, he is able to interrogate hisenvironment within a forward hemisphere ofabout ten feet. At ranges of a few feet or less,he is able to differentiate rectangular objectsfrom cylinders and sizes down to on the order ofone inch.

Daniel Kish operates “World Access for the Blind.” Google and Wikipedia are a good source for moreinformation on this capability. The wavelengths of sound in air are about 3.5 times longer than thosein water. The human auditory range is also limited to less than 15 kHz for practical purposes Therefore, the range and azimuth capabilities of human echolocation are considerably poorer thanthose of dolphins.

1.1.2 Overall vocal and auditory physiology of the dolphins

The literature of dolphin anatomy and physiology has been lacking in detail until 1995 or later. Theadvent of computer-aided tomography has led to rapid advances since that time, albeit based on only

Figure 1.1.1-4 Doppler frequency shift as a function ofrelative velocity in sea water.

5 September 200916

34Houser, D. Finneran, J. Carder, D. et al. (2004) Structural and functional imaging of bottlenose dolphin(Tursiops truncatus) cranial anatomy J Exp Biol vol 207, pp 3657-3665

a few examinations. Houser et al. provides valuable anatomical data but little discussion of how theparts play together34.

The physiology of hearing in dolphins and other members of Cetacea is built on the genericmammalian vocal and auditory systems. However, significant modifications and extensions havebeen introduced into the system. A prominent reorganization of the overall sensory architecture hasresulted in the introduction of a precision echolocation servo system (PES), or a precisionecholocation servo loop, much like the precision optical servo loop of vision in the higher mammals. Figure 1.1.2-1 illustrates these changes in block diagram form. The active echolocation systemrequires the close integration of the vocal system (nominally a stage 6 activity) and the auditorysensory system (stages 1-3), along with the expanded (stage 4a) signal manipulation capability, notinvolving the cerebral cortex, required to extract the additional information provided.

Dolphins are known to produce a wide range of acoustic signals in conjunction with the above system. Reynolds, et al (page 76) have provided a simple table of dolphin sounds. Table 1.1.1-1 provides anexpanded table relating more functional parameters. A much more comprehensive description of

Figure 1.1.2-1 Top level block diagram of vocal and auditory systems in dolphin. A new precision echolocationservo loop has been implemented by combining the active sound generation capability of the expanded vocal system(stage 6) with the wide frequency spectrum binaural auditory system (stages 0 through 4).

5 September 200917

35Herzing, D. (2000) Acoustics and social behavior of wild dolphins: Implications for a sound society InAu, W. Popper, A. & Fay, R. Op. Cit chapter 5

dolphin sounds is given by Herzing35. Even the Herzing description is light on the characteristics ofsignaling during echolocation and detailed target analysis. It does not address the physiology ofsignaling.

1.1.2.1 Acoustic signals used by the dolphin

Table 1.1.1-1Acoustic signals processed by dolphins

Frequency Range Sound Type Active/Passive Function

Coarse Long Range Capability

0.2–0.9 kHz Phase coherent listening Passive Source Location

4–20 kHz whistles Spatially coherent reflections Active Location, Ranging,& Doppler velocity

4-20 kHz barks, One way signaling Active Intra speciesyelps, rasps communications

Fine Short Range Capability

30–150 kHz clicks Spatially coherent reflections Active Location, Ranging &potential acousticimaging

The low frequency active system operates in a frequency range where the wavelength of the emittedsound is long with respect to the size of many food items. Furthermore, it is long with respect to thegeometry of the vocal apparatus within the dolphin. The resulting system is not very directionallyspecific. The source energy is distributed nearly omnidirectionally but the binaural capability of thereceiving system provides some directionality. It is used primarily for investigating the animalsenvironment (distance to surface, bottom characteristics and distance) and detecting large objects orgroups of objects. Similarly, the lowest frequency capability is limited primarily to the binauralreception of signals generated by external sources.

The high frequency active system operates in a range where the wavelength of the emitted sound issmall relative to potential food items. As will be discussed below, it is quite possible the dolphin isable to create an acoustic image of individual targets within the narrow beam of its field of view atthese frequencies.

5 September 200918

36Porter, M. Siderius, M. & Kuperman, W. ed. (2004) High Frequency Ocean Acoustics. NY: AmericanInstitute of Physics Preface

37Au, W. (1993) The Sonar of Dolphins. NY: Springer-Verlag pp 26-30

38Ketten, D. (2000) Cetacean ears In Au, W. Popper, A. & Fay, R. eds. Hearing by Whales and Dolphins.NY: Springer Chapter 2

39Supin, A. Popov, V. & Mass, A. (2001) The Sensory Physiology of Aquatic Mammals. Boston, MA:Kluwer Academic Publishers pg 181

Various authors have attempted to segregate the frequency spectrum of high frequency acoustics36. The division appears clear for the dolphins and most other members of Cetacea. The low frequency region includes frequencies below 200 Hz (generally restricted to the large whales). The highfrequency region includes the frequencies associated with the larynx, from 200 Hz to 30 KHz. Theultra high frequency region is associated with the nasal features between the larynx and the blowhole and extends from 30 kHz to higher values.

1.1.2.1 Modification of the non-neural auditory system

The changes in the non-neural portions of the auditory system have been so large as to escape cleardocumentation until at least 199337. Ketten has provided a major discussion of the morphology of theears of Cetacea but it does not address the physiology as specifically38. It is now generally recognizedthat the exterior ear (the pinna) and the auditory canal of the mammalian template are highlyatrophied and not used in Odontocetea. The remaining bridge between the hearing system of thetypical terrestrial mammal and the cetaceans are the sea lions (Classification: Suborder Pinnipedia,Family Otariidae, Subfamily Otariinae, Genus Zalophus, Species californianus). They are one of thefew members of their Suborder with conventional external ears and they use them when on land inthe conventional manner, for communications.

As late as 2001, Supin, Popov & Mass were still unable to express how odontocetes acquired theacoustic energy used in hearing39. They assert without conviction, “All hypotheses of functioning ofthe ear in cetaceans imply delivering the sound to the middle ear.” and “it remains debatable howsounds are delivered to the middle ear.” As will be shown below, both the “auditory canal and themiddle ear are non-functional in odontocete hearing.

1.1.2.1.1 Introduction to the outer ear of the dolphin

The most detailed data on the outer ears of the dolphin appear to be that of Ketten. She has focusedher early career on the anatomy and morphology of Cetacea, and primarily dolphins. She notes earlyin her writing of 2000 that the conventional pinna and auditory canal are not functional in cetaceanhearing. “In general, odontocete external canals are plugged with cellular debris and dense cerumen,becoming progressively narrower, and ending in a blind caecum that has no observable connectionwith the tympanic membrane or temporal bones.”

The function of the conventional outer ear is provided by an alternate external ear structure based onan acoustic lens concept rather than an acoustic horn. The acoustic lens is well-understoodtechnology in engineering. It is the specific configuration and density of all of the relevant structuresalong the inner jaw of the dolphin that remains poorly understood. This problem is complicated bythe apparent ability of the dolphins to change the exterior contour, and probably the interiorgeometry of their outer ears. Field studies have shown the two outer ears achieve a very broad

5 September 200919

40Ketten, D. (1998) Dolphin and bat sonar: convergence, divergence, or parallelism Fourth InternatBiosonar Conf pp 1-43

acceptance angle in the high frequency regime below 30 kHz but achieve a significantly narrowerforward projecting beam angle in the 120-150 kHz range.

As Ketten notes, “. . . the bulk of recent experimental and anatomical studies indicate specializedfatty tissues in the jaw region are the primary route for conveying sound to odontocete middle andinner ears.” Her 1998 work has provided additional details related to the arrangement of the fattychannels along the inner jaw40. Figure 1.1.2-2 reproduced from that work shows the level of detailavailable concerning the fatty channel material in bottlenose dolphins. Whether the regions of fatare employed in a single phased array or as separate energy collectors is not currently known.Further definition of these fat deposits will be found below.

1.1.2.1.2 Introduction to the outer-to-middle (or inner) ear interface in thedolphin

Au, writing in 2000, relied upon a caricature from McCormick (1970) to illustrate the middle ear ofthe bottlenose dolphin. The sketch and discussion are not definitive as to how the sound from theouter ear is transferred to the inner ear. He notes when discussing T. truncatus, “There is no directconnection between the tympanic membrane and the malleas.” It is surmised that the acousticenergy within the mandibular fat channel are delivered to the middle ear via a thin walled portion of

Figure 1.1.2-2 Three discrete lobes of highly differentiated fats have been identified, each oriented in a differentaxis, which may act in a tripartite sound collecting array in odontocetes. From Ketten, 1994 & 1998.

5 September 200920

the bulla itself.

Ketten approached the question differently. “In modern Cetacea, the ear bone consists of twoconnected bullae, properly called the tympano-periotic complex, that differ from temporal bonecomplexes of other mammals in form, construction, position, and, possibly overall function.” Shenotes, “There is considerable debate at the current time concerning the function of the middle earspace in cetacean hearing.”

If the energy was transferred to the malleus by one of the ligaments, an evolutionary variant of thesystem would be able to operate in two modes, as in the sea lion. In that variant, the acoustic energyfrom the auditory canal would be introduced by the malleus from the tympanic membrane. Theacoustic energy from any aquatic source would be introduced by motion of the external wall of thebulla and the appropriate ligament.

Since no impedance matching is required between the mandibular fat channel and the fluids of thevestibule associated with the inner ear, the acoustic energy may be delivered directly to thevestibular fluid where it can be applied to the cochlear partition of the inner ear. Under thisassumption, all of the bones of the middle ear are archaic and unused in the dolphin. The energy isdelivered directly to the large vestibule of the otherwise atrophied vestibular portion of the labyrinthsystem and then converted into a surface acoustic wave traveling along Hensen’s stripe on thetectorial membrane as in other mammalian inner ears.

Another feature of the middle ear concerns the thick vascularized mucosa, the corpus cavernosum,that lines the chamber. It is highly distensible and capable of filling the chamber (Ketten, 2000). While it is known that air fills the chamber in animals at the surface, the character of the chamberduring diving is not currently known.

Ketten notes, “The tympano-periotic complex resides outside the skull in an extensive peribullarcavity.” Comments occur in the literature suggesting the movement of the internal ears nearer theperiphery of the head in large animals would aid in the binaural location of passive sound sourcesand the reflections from illuminated targets. Such an assertion appears largely irrelevant since therelevant binaural distance is measured between the centers of the external ear collection areas andhas little to do with the location of the internal ears. This feature would need to be examined interms of the additional delay associated with the slow projection of neural signals from the internalears to the brain stem.

1.1.2.1.3 Introduction to the inner ear in the dolphin

Except for the greater degree of rigidity built into it, the inner ear is a conventional mammaliancochlea in most respects. Au presented considerable data in 1993 based on the investigations of theWever team in the 1970's. Its most important functional characteristic is the low rate of curvature ofHensen’s stripe in the basal region. This feature provides the high frequency performance, and thegood differential frequency performance in that region, of the dolphin cochlea. The number andarrangement of inner and outer hair cells are virtually identical to that in the human. Anothersignificant feature is the very large number of ganglion cells emanating from the spiral ganglia andforming the auditory nerve. The ganglia are about three times more numerous than in the human.

When discussing the inner ear, Ketten noted the minuteness of the vestibular system, associatedwith the inner ear, in dolphins. It appears the circular canals are minimally functional in thedolphins. As a result, the animal probably has little sense of up or down except as determined from

5 September 200921

41Johnson, C. (1986) Dolphin audition and echolocation capacities In Schusterman, R. Thomas, J. & WoodF. eds. Dolphin cognition and behavior: a comparative approach. Hillsdale, NJ: Erlbaum Assoc pp 115-136Chapter 5

the surface reverberation signals obtained from the echolocation system.

A significant difference in the inner ear of Cetacea compared to other mammals involves the outerear to inner ear interface. In Cetacea, the energy from the outer ear is applied to a different wall ofthe vestibule than in other mammals. Terrestrial mammals introduce sound energy into thevestibule through the oval window (and the associated stapes). In Cetacea, the stapes and ovalwindow have atrophied. The sound energy is introduced into the vestibule through the operculum,an acoustically pliable wall of the vestibule chamber as shown in Figure 1.1.2-3. The distancebetween the ears of the bottlenose dolphin is approximately the same as in humans, YO • 8 inches. However, the effective aperture of each external ear is much larger, XO • 3 inches.

1.1.2.2 The expansion of theauditory neural system

Figure 1.1.2-4 illustrates the major changes tothe neural system schematically as they applyto the auditory system. The major changesinvolve the outer ear (which has changedbeyond recognition from a morphologicalperspective), the major expansion of thelemniscus area and the geniculate nuclei(relative to their size in humans), and the closeintegration of the vocal and auditory systems inorder to achieve their synchronous operation.

It should be noted that the vestibular system is proportionately much smaller in dolphins than inhumans and other mammals of similar size. Its nearly vestigial size suggests that orientation withrespect to the gravity vector is much less important in the dolphin than its orientation with respect toits acoustic environment. This leaves the arrangement of the chambers immediately behind thestapes and oval window unclear. It will be assumed the common chamber, known as the vestibuleand normally found behind the oval window, is still there and it supports the transfer of longitudinalacoustic waves within the vestibule to surface acoustic waves on the surface of the tectorialmembrane (Fulton, 2006). Nachtigall (p. 96) has discussed the transfer of acoustic signals from thefatty channel of the outer ear directly into the vestibule without reliance upon the ossicle bones of themiddle ear. Johnson has also discussed schematically the transfer of energy to the fluids of the innerear by bone conduction41.

Figure 1.1.2-3 The outer/inner ear interface of thedolphin. Note the atrophy of the stapes/oval window. Sound enters the vestibule of the labyrinth from theLuneberg lenses forming the ears of dolphin (shownshaded) through the operculum (a different wall) of thevestibule and proceeds to the cochlea.

5 September 200922

Figure 1.1.2-4 The generic mammalian physiology of the auditory system applicable to the dolphin. The physicalproportions of these elements differ greatly from their counterparts in humans. The lateral lemniscus is almostvestigial in humans but massive in the dolphin. From Fulton, 2006.

5 September 200923

42Aroyan, J. McDonald, M. Webb, S. et al. (2000) Acoustic models of sound production and propagation InAu, W. Popper, A. & Fay, R. eds. Hearing by Whales and Dolphins. NY: Springer Chapter 10

1.1.2.3 The expansion and optimization of the phonation system

As Cranford noted (p. 144), the production of simultaneous whistles and clicks from the dolphinshave been reported many times over the years. However, only in the last five to ten years has itbecome clear how this is accomplished. The structure of the larynx and nasal passages of thebottlenose dolphin differ significantly from other species.

1.1.2.3.1 The composition of the melon in the dolphin

Morris (p. 370-377) has discussed the evolution of the head and jaw fats in odontocetes. The variationin the density of these fats and their approximate changes in formula are well documented at acoarse spatial level. However, the data is too coarse to define the beam forming capabilities of themelon in signal generation and the material along the lower jaw of these animals in signal reception.

1.1.2.3.2 The efficient generation of sound by Cetacea

Aroyan, et. al. have addressed the constraints on the generation of sound within Cetacea42. Both thehigh frequency regime of the bottlenose dolphin and the low frequency regime of the blue whale wereaddressed. The necessity of generating the low frequency acoustic signals within an air-filled cavityare stressed. The overall presentation stresses the very limited knowledge available at that timeconcerning sound generation in these animals.

Not addressed in the current literature are the optimal methods of generating the variousfrequencies used in dolphin echolocation. A conventional larynx is only capable of generatingfundamental frequencies up to about 1000 Hz. Above that level, the aural cavity is used to select thehigher harmonics of the fundamental when desired. To achieve fundamental acoustic frequenciesabove 5000 Hz using a stretched membrane is difficult because of the fragility of the resultingmembrane (note the frequently broken strings on a violin). At the high frequencies used forecholocation, a different method of sound generation is desirable (required).

The efficient generation of high power density single cycle pulses by Cetacea is not well understood atpresent. The conventional wisdom is largely conceptual. It has been that air is moved passed a set ofphonic lips to generate such a signal. The material characteristics of these lips has not beendescribed in detail. However, it is not likely they are calcified like teeth.

In an alternate mechanism, the dolphin could achieve high power efficiency in its highest frequencysound generators by using an impact to excite a liquid crystalline fluid contained within the bursae ofthe phonic lips. Because of their low propagation velocities, liquid crystalline materials are able togenerate quite high frequency oscillations within physically small spaces. The oscillatory energyoriginates in the material of the bursae rather than in the pneumatic spaces.

The liquid crystalline material will conform to the dimensions of its container. This allows thefundamental frequency of the oscillations to be varied by changing the dimensions of the bursaeunder muscular control.

The oscillations within the bursae can be transferred to the melon by employing intermediate

5 September 200924

materials with a gradation in propagation velocities until a velocity approximating that found in seawater is reached. Such gradations have been recorded in the melon as noted above.

1.1.2.3.3 Alternate methods of high frequency sound generation

There are several methods of sound generation that do not involve the flow of air past a pair of lips orthe slamming together of two hard surfaces. One involves a temporary reduction in pressure in arelatively large volume and its rapid recovery, typically described as an alveolar click. A secondinvolves a similar but more severe reduction in pressure resulting in a vacuum bubble in a gas or avapor bubble in a liquid that can collapse upon a return of higher surrounding pressure. This actionis related to cavitation.

A velaric sound results from the movements of the tongue. It is produced by the release of a closureat the front part of the tongue while the back of the tongue forms a second closure. The blade, tip orside of the tongue move down, releasing the front closure so that air rushes into the mouth,equalizing air pressure (ibid.: 247). The sounds produced by velaric airstream are called clicks.

A more promising mechanism for high power, high frequency sound generation is similar to theaction of the tongue and front pallette in humans when making a clicking sound. Rather thanslamming two surfaces together, the tongue is separated from the pallette quickly, resulting in a briefcavitation in the air within the oral cavity. If the tongue could move at a few hundred cycles persecond (well within the range of human vocal chords), it could generate a series of clicks very similarto those of the dolphin. Under this scenario, the phonic lips are actually a phonic tongue and pallette. Cavitation can generate very significant acoustic forces. Cavitation is known to cause the pitting ofship propellers. In this case the energy would be transferred to the bursae forming the functionalequivalent of a pallette. The biological nature of the tissue involved would allow the repair of anydamage due to cavitation over time.

The oscillations within the bursae can be transferred to the melon by employing intermediatematerials with a gradation in propagation velocities until a velocity approximating that found in seawater is reached. Such gradations have been recorded in the melon as noted above.

1.1.2.3.4 The generation of the high frequency sound fields

Rommel (Reynolds et al, p. 55) provided an early caricature of the fundamental modifications to thethroat of the dolphin. Although not of current interest, it did discuss several important features ofthe system. It is the oldest figure found that describes the complete isolation of the tracheal pathfrom the esophageal path. As a result, there is no need for an epiglottis. This separation allows theindependent operation of these two systems.

Ridgway has confirmed to this author (personal communication) that sound generated by thebottlenose emanates exclusively from its blowhole. Questions remain as to the precise source of thedifferent sounds generated by the bottlenose. Wesley (personal communication) has suggested to thisauthor that the larynx has become extraneous to both the whistle and click generation process. Hebelieves the phonic lips are responsible for both of these classifications of sound.

1.1.2.3.5 The potential generation of dual click sound fields

5 September 200925

43Johnson, C. (1986) Op. Cit. pg 126

44Evans, W. (1973) Echolocation by marine delphinids and one species of freshwater dolphin J Acoust SocAm vol 54, pp 191-199

45Bullock, T. & Gurevich, V. (1979) Soviet literature on the nervous system and psychobiology of cetaceaInternat Rev Neurobiol vol 21, pp 47-127

46Ridgway, S. & Carder, D. (1986) Nasal pressure and sound production in an echolocating white whale,Delphinapterus leucas In Nachtigall, P. & Moore, P. eds. Animal Sonar: Processes and Performances. NY:Plenum Press pp 53-60

47Cranford, T. (1986) The anatomy of acoustic structures in the spinner dolphin forehead as shown by X-raycomputed tomography and computer graphics In Nachtigall, P. & Moore, P. eds. Animal Sonar: Processesand Performances. NY: Plenum Press pp 67-77

48Cranford, T. (1996) Functional morphology and homology in the odontocete nasal complex: implicationsfor sound generation J Morphol vol 228, pp 223-285

Johnson discusses an important point43. “Evans44 showed that the beam pattern for different pulsescan vary greatly, and that T. truncatus can emit two simultaneously, one 20° to left of center and one20° to the right, separated in time by about 80 microseconds, with both appearing together at 0°” Ifconfirmed, this observation justifies a significantly different interpretation of the sound generationfacility of the bottlenose dolphin.

Bullock & Gurevich (p. 87) quote Reznikov as claiming the far field pattern of the high frequencyclick is rotated at high speed independently of any head motion45. The instrumentation to achievethe Evans and the Reznikov measurements with precision is not trivial.

Ridgway & Carder have noted two separate nasal passages begin immediately below the blowhole inthe white whale, Delphinapterus leucas46. They have also noted that if an open catheter is placed inthe blowhole, the animal is unable to generate significant sound intensity levels. This fact seems toconfirm the pneumatic resonance form of sound generation as opposed to a potential mechanicalresonance phenomenon.

Cranford has determined the shape of the melon in the spinner dolphin in considerable detail47. As inmost dolphins, the melon is not symmetrical where it interfaces with the nasal passages. Cranfordestimates the left branch (with its bursae) of the melon is rotated 50° down relative to the midsagittal plane.

The recent detailed imaging of the pneumatic system in bottlenose dolphins in-vivo by Cranford,Amundin & Norris has confirmed only one larynx but dual pneumatic passages beyond the larynxuntil near the blowhole48. At the 2000 Acoustical Society Convention, the Cranford team assertedtheir records confirm independent and/or simultaneous pulse operation of the sets of phonic lips. They also assert they observed whistles from only the left phonic lips.

1.1.2.3.6 An alternate schematic for the generation of the high frequencysound fields

The work of the above investigators leads to a different schematic of the sound generation system indolphins from that proposed earlier by Rommel. Figure 1.1.2-5 illustrates the currentunderstanding of the morphology of the dolphin pneumatic system. Not shown are the additional

5 September 200926

49Aroyan, J. (2002) Simplified models of the dolphin echolocation emission systemhttp://members.cruzio.com/~jaroyan/publications.html & http://www2.cruzio.com/~jaroyan/Dolphin%20Model%204.htm

multitude of air sacs surrounding portions of the melon and behind and below the spiracular cavities. These sacs provide effective acoustic mirrors that protect the brain case and the bulla containing thecochlea from the high energy acoustic energy delivered to the melon. They may also act to form theacoustic output beam in conjunction with the melon.

The dual channel character of the system is quite different from the preliminary description of thedolphin forehead presented in several forums by Aroyan49. His draft 2002 paper contains a mixtureof morphology and soundings of the bottlenose dolphin, and caricatures of sound generation for amember of the super family Delphinoidea that is not homologous with the bottlenose. It also differsfrom the concept employed here in suggesting the UHF source employs;1. A source mechanism of broadband pulses,2. A damped resonator, and3. A projector of the signal produced by the first two components.

In this work, it is suggested that the UHF system in T. truncatus consists of;1. Two high frequency resonators acting as sources,2. Two independently controlled envelope modulators of the Gabor type, and3. A projector of the signals produced by the above first two components.

In the proposed configuration, the spectral width of the pulses would be less than one-third of thecenter frequency and would not normally be considered broadband by a communications engineer.

5 September 200927

The schematic is somewhat simplified in that Morris and others describe the movement of air fromthe premaxillary air sac upward past the phonic lips (more properly the valvular flaps of the nasalplugs according to Morris) into an accessory sac and finally into the vestibular sac. Whether theaccessory sac is independent of the connecting spiracular cavity is difficult to discern from the

Figure 1.1.2-5 The dual channel operation of the sound generation system in the bottlenose dolphin. This schematicprovides an explanation of several observed properties of the dolphins sound generation system. First, it isolates thesound generation system completely from the digestive tract. It provides two distinct air columns between the larynxand the blowhole. The individual sets of phonic lips can excite the melon separately and generate far field patternsthat are significantly different (either simultaneously or sequentially. Note the angle between the plane containingthe pairs of phonic lips and the horizontal plane.

5 September 200928

50Mackay, R. (1986) Whale heads, magnetic resonance images, ray diagrams and tiny bubbles InNachtigall, P. & Moore, P. eds. Animal Sonar: Processes and Performances. NY: Plenum Press pg 82

51Johnson, C. (1986) Dolphin audition and echolocation capacities In Schusterman, R. Thomas, J. & WoodF. eds. Dolphin cognition and behavior: a comparative approach. Hillsdale, NJ: Erlbaum Assoc pg 120

published record.

While the generation of whistles and other sounds is associated with the larynx in this figure, someinvestigators believe the generation of whistles has migrated exclusively to the phonic lips. Additional data derived from sophisticated instrumentation (beyond the current behavioral evidence)will be required to confirm this point.

The arrangement of the elements in this schematic offers a much broader UHF (and possibly HF)capability than does a single channel system. These capabilities depend on the degree of separationof the two pneumatic channels and their arrangement immediately below the blowhole acting as aclosure. If the blowhole caps the two pneumatic channels individually, they would continue to bedescribable as independent systems. However, if the blowhole only closes above a common space,then the overall system could also include a mode of operation where the pneumatic channels wouldbe joined.

In either case, the parallel character of the phonic lips leads to a system capable of independent dualchannel operation. This ability, at least in the generation of UHF clicks appears well documentedand well supported by the detailed geometry of the melon. It leads to the generation of twoindependent forward projecting beams separated by a distinct angle as reported above. These beamsmay occur simultaneously or sequentially. Their generation is compatible with a distinct spatialpolarization that may appear to be due to a rotating beam when measured with less than optimuminstrumentation. The spatial polarization may lead to the reporting of a distinct temporal phaserelationship between the two beams under these conditions.

By manipulating the volume of air in the air sacs and the volume of the lungs, the air capturedwithin the animal can be circulated back and forth repeatedly as desired. Thus, the air in the leftand right tracheal paths can be driven past a constricting larynx or the constricting nasal plug. Thehatched area represents the boney character of much of the wall of the tracheal airway in dolphins. The nasal plug is shown as a specialized portion of the melon. This schematic would support theconcept of dual sound sources within the trachea of dolphins. The larynx would produce soundswhistles within the frequency range and of the character normally associated with mammals of thissize. The nasal plug would generate the higher frequency sounds associated with the preciseecholocation facility of the dolphin. Mackay has confirmed the larynx is not the source of the highfrequency sound energy based on magnetic resonance imaging in-vivo and during high frequencyphonation50.

Johnson has provided a schematic of the various air-filled spaces in the body of the dolphin51. Thefigure generally follows that shown here. It makes no reference to the vestigial meatus and pinnawhen discussing the dolphin auditory system.

Bullock & Gurevich (p. 74) provide a variety of details concerning the muscles and neurons associatedwith the air sacs, the spiracular cavity and the tissue near the blowhole. They also support theassertion by Marjov & Tarchevskaya (1978) that the dolphin can make a variety of sounds duringexhalation and inhalation as well as during the apneustic plateau.

5 September 200929

52Cranford, T. (2000) Op. Cit.

Cranford has performed an extensive investigation of the nasal spaces in odontocetes seeking thesource of high frequency sounds52. However, the precise character of the blowhole closing could notbe found in that study. Cranford does address the limited ability of the muscles (even selectedmuscles) to operate at rates above 800 actions per second. This led to a study of friction-based,cavitation-based and pneumatic mechanisms. His conclusion is that the source of high frequencyvibrations is analogous to those generated in a brass wind instrument. He focuses on the importanceof the vibrations existing in the structure of the phonic lips. This feature avoids the necessity ofcoupling the vibrations of the air column into the tissue of the animal before projection into the waterenvironment. He supports his assertions with the results of the first reported set of high speed videoendoscopy recordings. The high frequency clicks are formed by one cycle of the explosive parting andclosing of the phonic lips. The waveform within the envelope of the explosion is apparently due to themechanical properties of the lips (plug, etc.). His conclusion is that the dolphin can generate twoindependent acoustic signals using the two nasal passages (or potentially two locations along onepassage). As others have noted, the sound generation complex on the right is approximately twicethe size of that on the left in T. truncatus. Cranford also notes the potential difference in frequencybetween the simultaneous clicks from the same animal.

The generation of two distinct UHF beams of acoustic energy leads to the description of this part ofthe vocal system as bivocal or biphonic in analogy to the binaural operation of the two ears.

In this description, the temporal phase relationship between the two acoustic sources is an entirelydifferent subject than the phase relationship between the signals generated within the outer and/orinner ears. The phase relationship between the two radiators could significantly affect the reflectedsignal from a target.

Cranford also noted another intriguing endoscopic observation in the dolphin. “the dynamic parallelfurrows that form in the surface of the nasal mucosa, between the nasal plugs and the phonic lips. The direction and depth of the furrows can be changed instantaneously, apparently by muscleaction.” This work suggests the mucosa is acting as a liquid crystal and the furrows are surfaceacoustic waves dependent on the size of the borders of the liquid crystal.

As a group, the above observations provide a strong foundation for the independent dual channelcapability of the phonation system of the dolphin.

1.1.2.3.7 Acoustic propagation within the melon and other cavities of thedolphin head

As noted by Cranford (p. 135), any study of signal generation would be incomplete without aconsideration of the effect on signal propagation of the various regions of different density within thehead of the dolphin. Not only are the densities of the various tissues and bones of the headsignificant, but the myriad air spaces represent near-perfect mirrors to the transmission of acousticenergy within the soft tissue of the head. He discusses these passages in reasonable detail but notesthat their dynamic characteristics have prevented any definite conclusions to date (p. 123). Anysignificant ray tracing applicable to the head of the dolphin must take into consideration theinstantaneous configuration of these air spaces. Cranford’s concluding remarks (p. 141-145)concerning future directions for research are very useful to the research community.

1.1.2.3.8 Pulse repetition rates in high frequency echolocation

5 September 200930

53Ridgway, S. (1986) Physiological observations on dolphin brains In Schusterman, R. Thomas, J. & WoodF. eds. Dolphin cognition and behavior: a comparative approach. Hillsdale, NJ: Erlbaum Assoc pp 115-136Chapter 2

54Fulton, J. (2006b) The Electrochemistry of the Neuron. Sec. 6.3.5 www.sightresearch.net/neuron/pdf/6Electrophysiol.pdf

Odontocetes are known to vary the pulse repetition rate during echolocation so as to prevent multiplepulses to be propagating in the water simultaneously. At long range, the pulse rate is necessarilylow. Pulses may be separated by 50 msec or more (20 pps). At close range, the pulses can be as closetogether as 1.5 msec (660 pps). Avoiding simultaneous pulses in the medium is of great advantage insimplifying the separation of multiple targets and the removal of signal clutter due to reverberation.

Cranford showed concern (p. 113) that subsequent pulses were frequently generated before thereturn signals were completely processed. In good radar and sonar system design, there is no needfor the returning pulses to be processed prior to the generation of the next pulse. Once the returnshave reached the neural system, they are isolated from the environment and can be processed “off-line” using a pipeline approach as long as the temporal integrity of the signals is maintained. Such apipeline approach may even involve multiple parallel pipelines as long as temporal integrity ismaintained.

It should be noted that 660 pps is near the limit of the firing rate of specialized neurons, such asthose found in the precision optical system of mammalian vision, the binaural neural circuits ofmammals, and probably in the precision echolocation system of the dolphin.

The unique characteristics of click trains in dolphins have spawned a number of specialized terms. The pulse rate of a rapid train of pulses can be heard by humans. This tone, defined by the pulserate, has been labeled the “time-separation pitch.” Cranford has noted the systematic variation inthe number of pulses in a pulse train among different species of odontocete. He suggests those pulsetrains of less than seven pulses be labeled oligocyclic. Longer trains would be described as polycyclic.It is too early to associate the other characteristics of a pulse train described by Cranford (p. 115)with these two labels unequivocally.

1.1.2.4 The unique neural circuits of echolocation

Several investigators have noted the unusually large size of some of the axons connecting the spiralganglia to the CNS. Ridgway has considered the significance of these large axons primarily withrespect to propagation delay times53. However, such axons are also capable of propagating widerbandwidth signals54. The size of these axons, and the identification of a unique operating regimeextending from 0.2 to 0.9 kHz by Reynolds, suggest these axons are associated with phase-coherentprocessing of the signals delivered to the passive source location circuits, and probably the activeecholocation circuits of the CNS. Data is beginning to accumulate suggesting the passive locationcircuits are located in the inferior colliculus and the active location circuits are located in the laterallemniscus of the paleocortex.

Ridgway has noted that “Bullock & Ridgway suggested that the dolphin brain might possess twoseparate auditory systems, one specialized for ultra-brief, fast-rising sounds like echolocation clickswhile th other was specialized for longer, slower rising sound like dolphin whistles.”

1.1.3 Overview of signaling in dolphins

5 September 200931

55Caldwell, M. Caldwell, D. & Tyack, P. (1990) Review of the signature-whistle hypothesis for the Atlanticbottlenose dolphin In Leatherwood, S. & Reeves, R. eds. The Bottlenose Dolphin. NY: Academic Press Chap 10

56Aroyan, J. (2002) Simplified models of the dolphin echolocation emission system. http://www2.cruzio.com/~jaroyan/Dolphin%20Model%204.htm

It appears generally accepted now that the dolphins make a variety of sounds for purposes of bothcommunications and echolocation. The range of the sounds produced by odontocetes is so wide, theyhave been informally labeled “sea canaries.” They may also generate high frequency signals for thepurposes of stunning their prey. However, this latter subject is beyond the scope of this work. Bullock & Gurevich (and many others) have noted the wide variety of signals generated by dolphins. However, it has been difficult to categorize these sound in a non-overlapping matrix. Bullock &Gurevich described the possible sounds studied by Markov, et. al. and their permutations andcombinations using thirty-one alphabetical symbols (fortunately Russian contains more letters thanEnglish) and five combinatorial rules. They did not attempt to associate semantic rules with thesecombinations. Others have described up to nine combinatorial rules. Caldwell et al.published largetables of the potential sounds from the bottlenose in captivity (which may not be complete because ofthis constraint)55.

This work will only address the two major classes of dolphin sounds, clicks and whistles. Only thosewhistles having characteristics that might be associated with echolocation operations will beexplored.

One of the obvious problems with studying the morphology of the dolphins is the apparent ability ofthe animals to dynamically change the physical shape of many of the elements employed inecholocation. The artistic and photographic figures in Reynolds suggest the wide variety of externalhead shapes encountered in dolphins under different conditions. As a result of this situation, mostschematic drawings of the head of the dolphin in the technical literature must te taken as exemplarybut not definitive of the class, or even the particular animal under different conditions.

Recent activity in dolphin research has been centered on understanding the detailed morphology ofthe echolocation apparatus in dolphins. Separate teams are exploring both the signal radiation andthe signal reception systems. Aroyan has provided a useful pedagogical paper (in incomplete draftform) concerning the signal generation aspects for dolphins56. His modeling should not be taken asthe precise model of the system but of a potential model. His reliance on a parabolic reflector basedon the shape of the skull is probably over-constrained (particularly when compared to his figures 1 &2). Alternate descriptions of the source that do not involve a broad band noise source are worthy ofconsideration. His figures 9 and 10 may only apply to a subset of Cetacea with only one functionalsource of high frequency pulses. There is considerable data supporting two active sources of highfrequency pulses in T. truncatus. A point Aroyan does illustrate is the wide variety of waveformsgenerated by dolphins. It is not likely all of these signals can be generated by one geometrically fixedsource. As in human speech, the variety of signal waveforms suggests a dynamic ability to vary thespatial geometry of both the source and beam forming elements of the system. Such dynamics makesprecise description of the system difficult.

Potter & Taylor have provided an interesting Power Point presentation on a novel hypothesis that

5 September 200932

57Potter, J. & Taylor, E. (2001) On novel reception models for bottlenose dolphin echolocation. www.arl.nus.edu.sg/objects/paper_ver1.pdf Proc Inst Acoust vol 23, pp 103-112

58Potter, J. & Taylor, E. (2002) On novel reception models for Bottlenose dolphin echolocation Http://www.arl.nus.edu.sg/objects/dolphin.ppt

59Hoffmann-Kuhnt, M. Chitre, M. & Potter, J. (Undated) “Ghosts in the Image: – aliasing problems withincoherent synthetic aperture using a sparse array. www.arl.nus.edu.sg/objects/0_327.pdf

60Houser, D. Finneran, J. et al. (2004) Structural and functional imaging of bottlenose dolphin (Tursiopstruncatus) cranial anatomy J Exp Biol vol 207, pp 3657-3665

the individual teeth of the dolphin play a major role in the reception of acoustic energy57,58. Hoffmann-Kuhnt, Chitre & Potter have continued the study of this hypothesis and presented a paperdiscussing the application of sparse array signal reception applicable to the dolphin59. The papermakes a variety of assumptions concerning the nature of the receiver geometry but does not discussthe simple binaural character or the neural data processing associated with the hearing system. Their discussion of signal processing appears to relate more to their concepts of incoherent syntheticarray signal processing than to the character of the receiving system in dolphins. This work willfocus on simpler receiving configurations.

Houser, et. al. have continued a program of imaging the head of the dolphin using modern medicalimaging techniques and included a thorough bibliography in their report60. An important observationwas made in this paper. It is critically important that the cochlea of the dolphin be isolated from thehigh intensity (estimated 60 watt peak level) acoustic generator located less than twenty centimetersaway. It is important that air sacs be provided to act as acoustic mirrors for two reasons. First, toaid in effective beam formation, and second, to shield the bulla containing the cochlea. Houser, et. al.noted, “Air spaces directly abutted the tympano-periotic complex such that a bone-air interfaceexisted. Coverage was most complete on the dorsal medial and posterior surfaces of the tympano-periotic complex, with the dorsal surface being almost completely covered by a layer of air.”

1.1.3.1 The external anatomy and beam forming capability of the outer ears

The outer ear of the dolphin has been so greatly modified that it is hardly recognizable within theframework of conventional morphology. Functionally, the morphological horn type energy collectorsat each side of the head have been replaced by hydrodynamically-streamlined surface-mounted lenstype energy collectors below each jaw. The conventional auditory canal (meatus) and pinna of thedolphin is considered vestigial and no longer associated with hearing.

For the remainder of this discussion, the “outer ear” of the dolphin will refer to these two lens-typereceiving devices and not the vestigial elements of the horn-type receivers.

While many artistic and photographic renditions of the front quarter profile of the dolphin exist, it isdifficult to find one describing the outer ear appropriately. Figure 1.1.3-1 shows a modified drawing,credited to Marya Willis-Glowka (Reynolds, p. 20), showing the location of the forward facing lens ofthe outer ear and the location of the vestigial auditory canal (now probably functioning as a staticpressure port). The internal geometry of the “outer ear” forms a forward-looking beam at highfrequencies but a more omnidirectional beam at low frequencies.

5 September 200933

61Caldwell, D. & Caldwell, M. (1972) The World of the Bottlenosed Dolphin. NY: J. B. Lippincott pg 108

62Floyd, R. (1986) Biosonar signal processing applications In Nachtigall, P. & Moore, P. eds. AnimalSonar: Processes and Performances. NY: Plenum Press pg 779

The forward-looking aspect of the hearing receivers (outer ears) is shown more clearly in Figure1.1.3-2, modified from a photograph in Caldwell & Caldwell61. The variations in the prominence ofthe external ears in different renderings of the bottlenose dolphin suggest the shape of these featuresmay be changeable at will by the animal. Kassewitz has noted the great and rapid motility of thetissue in both the melon area and the lower jaw area of the dolphin (personal communication). Thismotility suggests the steerability of both the UHF sound generation system as well as the steerabilityof the reception system. The configuration of the radiating and reception elements of theecholocation system is similar to that described by Floyd in Nachtigall & Moore62.

Figure 1.1.3-1 The outer ear of the bottlenose dolphin. The lens system of the outer ear is forward looking at highauditory frequencies but nearly omnidirectional at low frequencies. The vestigial auditory canal (absent any pinna)may function as a static pressure port largely independent of the hearing system. Modified from Marya Willis-Glowka.

5 September 200934

63Ketten, D. (2000) Cetacean ears In Au, W. Popper, A. & Fay, R. eds. Op. Cit. pg 46

Before beginning to analyze the physiology of the echolocation system in dolphins, it is veryimportant to note three unique conditions. First, the differences in acoustic impedance between airand both water and the soft tissue of the dolphin are so great that the air-tissue interfaces within thehead of the dolphin act as acoustic mirrors63. Second, the wavelengths of the signals generated byand received by the dolphin are frequently long relative to the dimensions of the animal and itsrelevant parts. Thus, the energy propagating internally and within about one-meter or less of theanimal must be analyzed under near field conditions. This condition means that simple geometricray tracing can only provide an approximation of what is actually occurring. A full diffraction-basedray tracing activity is required to obtain accurate results. Harris has reviewed the near field and far-

Figure 1.1.3-2 A front view of the bottlenose dolphin showing the forward looking lens-type external ears and thegeneral location of the melon used for echolocation (dotted overlays) . The caption of the Caldwell & Caldwellfigure hint that the eyes and possibly the eye sockets of the dolphin may rotate to support their apparently goodstereoscopic vision. The eyes are seen more clearly in the original art. Modified from Caldwell & Caldwell, 1972.

5 September 200935

64Harris, G. (1964) Considerations on the physics of sound production by fishes In Tavolga, W. ed. MarineBioacoustics. NY: Pergamon Press pp 233-247

65Mackay, R. (1986) Whale heads, magnetic resonance images, ray diagrams and tiny bubbles InNachtigall, P. & Moore, P. eds. Animal Sonar: Processes and Performances. NY: Plenum Press pp 79-86

field effects64. Mackay has also addressed the near field situation and suggested solutions usingdifferential equations are required65.

Third, investigators have frequently discussed the conduction of sound within the boney structure ofthe skull in mammals. In the dolphins, the auditory bulla has migrated to a point outside of the skulland it is not attached to the skull by bone. In addition, most of the potential sound conductingpathways with the dolphin are made of soft tissue. Therefore, the conventional physiological termbone conduction is probably best replaced by tissue conduction when speaking of dolphins, andprobably other members of Cetacea (as suggested by Reppening in 1972). However, such a change interms is unlikely in the short run.

5 September 200936

By employing gradient index materials in the fatty channel connecting these energy collectors to thebulla, excellent forward looking acoustic performance has been achieved at frequencies near 120 kHz. The field patterns become more omnidirectional as the frequency is reduced. The data is very sparseand generally obtained as a composite based on binaural hearing. Analyzing the data is awkwardbecause of the limited data available on the geometry of the fatty channels behind the aperture. Bydecomposing the binaural field pattern, a first order estimate of the single-ear performance can bemade. Figure 1.1.3-3 shows how these patterns might look for the two ears individually. Theasymmetry of the patterns would suggest the presence of an acoustic ground plane within the beamforming structure. This structure is probably the bone of the lower mandible. If the mandible isacting as a ground plane, it would explain the assertions that the mandible and teeth are areas ofhigh auditory sensitivity in the dolphin when excited by a contact source of acoustic energy. Suchexcitation would not suggest the mandible is an active part of the beam forming elements of thereceiver and would not be suggestive of the far field performance of the mandible.

5 September 200937

The overall field pattern due to one ear does suggest the fatty channel forms a conduit containing oneor more reflective elements and/or refractive elements. The mandible would be an example of areflective material. A variable index in the fatty channel itself would be an example of a refractive

Figure 1.1.3-3 Estimated field patterns of the auditory receivers of the bottlenose dolphin at 120 kHz. Heavy solidlines; left ear. Broken lines, right ear. The sum of these two patterns would result in a single pattern with a principlelobe along the centerline of the animal. From data of Au & Moore, 1984.

5 September 200938

material. This fatty material is known to be heterogeneous (Morris, p. 369).

The small bones of the middle ear are not used in the hearing of Cetaceans. Ridgway notes thestiffness of the ossicles of the middle ear and the malleus does not attach directly to the tympanicmembrane. The basilar membrane is within a mm or two of the length of the human membrane butit is more firmly supported at the basal terminus. The number of inner and outer hair cells appearsto be comparable with the human but the number of ganglion cells in the spiral ganglia is muchgreater. Where the ratio of ganglion cells to hair cells is about 2:1 in humans, it is about 5:1 in thebottlenose dolphin.

The dolphin lacks a pinna and the external meatus has apparently become a vestige as will bediscussed below.

Several current members of the dolphin family are described graphically by Au on page 136 andperformance-wise on page 134. Chapter 2 describes the hearing system and Chapter 5 describes theactive sound generation system. Chapter 10 describes the estimated signal processing capability ofthe species without the benefit of significant physiological data. The signal processing capability isderived primarily from performance data collected under a variety of conditions but without a goodmodel of the biological systems involved.

The high frequency acoustic performance capability of the animal is very completely characterized inAu. However, the physiology of the individual acoustic elements within the head is not well detailedat this time. The discussions in Au of the beam forming capability of the dolphin and the signalprocessing capability (in analogy to other radars and sonars) are both elementary and should not berelied upon at the detail level. His section 10.3.5 agrees with this statement. Chapter 10 exploredthree signal processing models; an envelope detection model, a matched-filter model and aspectrogram correlation model. The comments concerning the time required to compute thespectrogram are largely irrelevant if the dolphin uses a spatial signal separator and spatial signalcorrelator as expected based on its generic mammalian hearing capabilities.

Au makes a number of important distinctions between the capabilities of bats and dolphins. Hegenerally limits dolphins to short pulse (wide band) signal generation while bats are known to emitlonger pulse (narrower band) signals when required to achieve a Doppler radar capability. Whiledolphins are known to use (and are trained to respond with) long duration whistles, it may be thatthese are generated by the larynx and not related to their sonar capabilities. On the other hand, itappears they are trained to whistle while underwater which suggests the inherent capability ofachieving Doppler sonar performance using this lower frequency sound source (typically 5 to 30 kHzlasting up to several seconds; Au, page 77).

The term broadband should be used sparingly when discussing the clicks of odontocetes. Theenvelope of the clicks in most species are shaped to resemble a Gabor function. As a result, theoverall signal consists of a center frequency and two major sidebands plus only a fewsignificant harmonics of those sidebands. The resulting frequency spectrum may have a Q ofless than four, but it is far from a flat spectrum over a considerable bandwidth.

The comparison of the dolphins capabilities and those of a relatively simple radar system areprobably not useful. The performance of the dolphin system is probably better than, or at least asgood as, state-of-the-art single pseudo-noise pulse radars used in modern stealth weapon systems. While the broad-bin histograms in section 7.2 of Au (1993) are illustrative, they do not show theamplitude and phase relationships between individual components of the sideband information that

5 September 200939

can be used effectively to recover additional information. Au makes note of the potential of the“elementary signal” defined by Gabor, as the basic function in signal decomposition rather than theconventional sinusoidal function in Fourier analysis. He also presents information on multiple pulsecorrelation and rippled waveforms in chapter 4. These concepts suggest the capability of the dolphinsystem is quite advanced relative to, or equals the performance of, similar man-made systems.

Au suggests on page 137 that acoustic power output levels of 60 watts are achievable in somedolphins.

In light of the generic performance of mammalian hearing, it is likely that the dolphin uses both theenvelope detection model and the spectrogram correlation models simultaneously. This will beexplored in conjunction with the proposed dolphin echolocation schematic in a following section.

1.1.3.2 The external anatomy and beam forming capability of highfrequency vocalization

1.1.3.3 Typical signaling waveforms of the dolphin

Bullock & Gurevich have provided a useful discussion of dolphin sounds (p. 74-81). A problem hasbeen the infrequent separation of the sounds below 30 kHz into purpose-oriented classes. From asystem perspective, the unique swept continuous tone signal should be separated from all of the otherquacks, croaks, screams, barks, etc. The timing and other characteristics of this swept waveformclearly indicate it is part of a continuous tone sonar system probably exhibiting a Doppler capability. The other sounds do not suggest such a capability.

It is proposed that all of the sounds generated by the bottlenose dolphin must be separated into threedistinct classes; the low frequency (LF) communications sounds, the high frequency (HF) sweptfrequency sonar sounds, and the ultra high frequency (UHF) pulse sonar sounds. Such a separationprovides a different interpretation than the comments of Markov, et. al, and Tomilin in Bullock &Gurevich (p. 76-77). When ranging, the HF whistles are necessarily monotonous because it is theirecho that carries information, not the emitted whistle. On the other hand, the LF sounds containinformation worthy of recovery by other members of the species. These sounds will be consideredelements of the dolphin language, described here globally as Dolphinia. Dolphinia is likely to containa great many dialects due to the separation of various herds by great distances.

Bullock & Gurevich (p. 87) reference Ayrapet’yants, et. al. as describing the social communicationsrange of the dolphin as 5-10 km.

1.1.3.3.1 Click-type waveforms

The literature contains a great many casual statements concerning the signal parameters associatedwith dolphins. The term broadband is frequently associated with the high frequency signaling ofdolphins. Figure 1.1.3-4 shows that the analogy is not necessarily appropriate. In this figure, thepulse transmitted at about 130 kHz occupies a bandwidth of less than one-third of the centerfrequency. Such a signal is generally described as a narrow band signal in engineering disciplines. Thus, the dolphin is capable of generating high power and narrowband pulse signals within a broaderrange of frequencies (80-150 kHz). This range is still less than 2:1.

5 September 200940

Figure 1.1.3-4 Temporal and spectral characteristics of ahigh frequency click. From Tyack & Clark, 2000.

5 September 200941

Figure 1.1.3-5 shows a click waveform measured repeatedly by Lemerande at San Diego. It shows avery good quality single cycle waveform at a nominal 78 kHz. This is truly a narrow band signal withsome harmonic content. This click would allow much more precise range and range ratedetermination (much smaller standard deviations) than possible with the above click from Tyack &Clark.

1.1.3.3.2 Whistle-type waveforms

Many of the following waveforms were recorded under recording system limited conditions withfrequencies typically limited to the human audio range. Only recently has Kassewitz implemented asingle system that can record from a few hundred cycles to 150 kHz (Section 1.7.1).

The spectral waveforms representing dolphin sounds are extensive but not easily comparable. Therange of sounds produced is great and the variation in the sequence of these sounds is alsoconsiderable. Figure 1.1.3-6 reproduces a typical low frequency spectrum presented by Sayigh(Reynolds, p. 77). This figure shows considerable harmonic content within a single emission.

The low frequency “whistle” sounds are usually described (Cranford, 2000, p. 111) as “generallynarrow bandwidth, often frequency modulated, sounds that commonly last from half a second to afew seconds and may have harmonic structure. Such whistles have been observed at depths of 300

Figure 1.1.3-5 A nearly optimal click with a fundamental frequency of 78 kHz and some harmonic content(particularly in the after pulse). Energy beyond 230 microseconds is believed to be reverberations within the skull ofthe animal. From Lemerande, 2002.

5 September 200942

66Houser, D. Finneran, J. et al. (2004) Structural and functional imaging of bottlenose dolphin (Tursiopstruncatus) cranial anatomy J Exp Biol vol 207, pp 3657-3665

meters, but with changed duration and frequency composition.”

The ability of Cetacea to make significant sounds at great depth (an area not generally visitedby dolphins) remains an area of research. Cetacea can be described as “breath holders.” As aresult, the air in their pneumatic passages is greatly compressed when diving, to the point thattheir rib cages are designed to collapse at depth. As a result, the available volume of air atdepth is negligible compared with the surface volume. There has been speculation that nearlyall of the air from the lungs is transferred to the nasal sacs to allow continued echolocation atgreat depths66. In the case of the bottlenose dolphin, soundings at depths exceeding 200 metershave been reported.

Figure 1.1.3-6 A nominal high frequency (HF) signal spectrogram generated by dolphin. From Sayigh in Reynoldset al, 2000.

5 September 200943

67Caldwell, M. Caldwell, D. & Tyack, P. (1990) Op. Cit.

Figure 1.1.3-7 shows an alternative and more extensive set of spectrograms from Caldwell et al67. These recordings display almost no harmonic content. The reason may relate to equipmentcapability. Only a few first harmonics (below 12 kHz) of low frequency components are shown. Theynote the difference in characteristics of the low frequency signals from two different juvenile males. Each individual sequence was labeled a loop in their presentation. They note the repetitive nature ofthe patterns in each loop. However, they also note the frequent occurrence of an introductory orterminal loop that differs from the central loops (sequence at lower right). The difference in the loopstructure between animals provides a significant interference rejection capability when severalindividuals are hunting in a single pack. Caldwell et al, provide considerable information concerningthe variation and stereotypical characteristics of whistles from specific dolphins.

It is very likely that the harmonic content (which may be controllable) of the emitted signals plays asignificant role in determining the radial dimension and the internal texture of targets. Theharmonic content would be quite useful in distinguishing a real fish from a metal imitation placed inthe water by a researcher. For larger specimens, it could help distinguish a real target from aStyrofoam impersonation of uniform density.

5 September 200944

68Caldwell, M. & Caldwell, D. (1979) The whistle of the Atlantic Bottlenosed Dolphin (Tursiopstruncatus)–Ontogeny In Winn, H. & Olla, B. eds. Behavior of Marine Mammals, Volume 3 Cetaceans. NY: Plenum Press Chap. 11

Caldwell & Caldwell had earlier described the whistles of 126 infant-to-young Atlantic BottlenoseDolphins during their development68.

1.1.3.3.3 Truncatus waveforms recorded in the Sado estuary of Spain

Santos, et. al. have identified a wide range of whistles recorded in the Sado estuary near Lisbon

Figure 1.1.3-7 A collage of LF spectrograms from two Atlantic bottlenose dolphins. The frequencies are below 20kHz. Left; one to six loops of a juvenile male. Right; one to six loops of a different juvenile male. Note thelabeling, A, B, B, B, C at lower right showing variations in the loops of one sequence. From Caldwell et al, 1990.

5 September 200945

69Santos, M. Caporin, G. et. al. (1989) Acoustic behaviour in a local population of bottlenose dolphins InThomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pp 585-598

Portugal from a resident herd of bottlenose dolphins69. They attribute these whistles to socialcommunications as much as to echolocation. Figure 1.1.3-8 reproduces their two tables.

5 September 200946

Figure 1.1.3-8 Characteristics and occurrences of whistles in Sado estuary. The contours suggest eithercommunications or personal signatures as much as they suggest echolocation. From Santos, et. al., 1989.

5 September 200947

70Markov, V. & Ostrovskaya, V. (1989) Organization of communication system in Tursiops truncatusmontagu In Thomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pp 599-622

71Tyack, P. L. (1986). Whistle repertoires of two bottlenosed dolphins, Tursiops truncatus: mimicry ofsignature whistles? Behavioral Ecology and Sociobiology vol 18, pp 251-257

72Hickey, R. (2005) Comparison of Whistle Repertoire and Characteristics Between Cardigan Bay andShannon Estuary Populations of Bottlenose Dolphins (Tursiops truncatus): With Implications for Passiveand Active Survey techniques In fulfillment of a Master’s Degree, University of Wales (Bangor), GB Copyin WhDolphin folder

The Santos team recorded these signals from a boat trailing a herd by about 50 meters. Thus all ofthe sounds were recorded from a posterior location relative to the emitters of the dolphins. Signalsup to 24 kHz, but most up to 8 kHz, were recorded. A variety of spectrograms were provided basedon these recorded HF band signals. The variety of whistles in Table 1 and their occurrences in Table2 suggest several potential purposes;

< swept frequency echolocation with signals coded to specific animals,<signature whistles identifying individuals and their location within the herd, and<whistles used in intraspecies communications and not specific to an individual.

The animals were frequently found to form multiple herds traveling separately in the large bay. As aresult, whistles identifying the location of individual animals at significant distances would not beunexpected. Similarly, coded swept frequency signals would allow simultaneous echo ranging bymultiple animals with interference rejection using matched filter techniques.

Whether such signals are used in communications has been studied for years with less thandefinitive results. Markov & Ostrovskaya discuss the possibilities using caricature waveforms inconsiderable detail70. They review the history of communications studies of dolphins up to 1989. Tyack presented a repertoire of dolphin whistles in an obscure journal in 198671. There has beenlittle work since then on learning to understand any native dolphin language. Most of the recentactivity has involved interspecies communications.

1.1.3.3.4 Truncatus waveforms & spectra from the Irish Sea & Shannonestuary

Hickey studied the sounds, in the frequency range from 2.5 to 25 kHz, of several resident groups ofbottlenose dolphins72. The recorded waveforms were similar to those recorded in Spain. Hickey didnot cite the work of Santo et al. or the text by Thomas & Kastelein. He did cite similar work in theSado estuary reported by Harzen in 1998. Figure 1.1.3-9 reproduces Hickey’s figure 3.2. The toprow illustrates the basic forms used to create the designations used in the figure. Letter designationsare cascaded to describe the shape of a complete whistle. Hickey provided both his stylizedwaveforms and accompanying raw spectrograms. He did not address echolocation pulses in hismaterial.

Multiple statistical tests were applied to the data by Hickey to differentiate between different herds. He attributed most of the sounds to periods when the animals were foraging for food. Hickey alsodiscussed a variety of trends in the use of whistles by dolphins. He traced the designation ofpersonal, or signature, whistles and shared, or variant, whistles. The shortness of the majority ofthese whistles was highlighted. He also discussed the development of dialects between herds and

5 September 200948

73Taruski, A. (1979) The whistle repertoire of the North Atlantic pilot whale (Globicephala melaena) InWinn, H. & Olla, B. eds. Behavior of Marine Mammals, Volume 3 Cetaceans. NY: Plenum Press Chapter10 pg 347

74Jones, H. (2005) The Dolphin Defender. Video on PBS, Nature series

their relative isolation.

A problem with the Hickey notation is it does not include trills with harmonics present. It isproposed to employ a trailing numeric to indicate the number of harmonics present in a whistle. Anexample would be an A5 to designate the presence of the fundamental and harmonics through thefifth. The trailing numeric could be a superscript but that would complicate email messages.

The conflict between the notation of Santos andHickey is significant. Both differ from thenotation of Taruski for the pilot whale73. Markov & Ostrovskaya have presented anentirely different notation. A researcher mustbe specific as to the terminology followed orintroduced. These conflicts highlight two majorproblems. First, the authors may be speakingof different languages. There is no reason thata pilot whale should be speaking the samelanguage as a bottlenose dolphin. It appears ataxonomy of languages for the marinemammals is needed. While Lilly suggesteddelphinese for the bottlenose dolphin (page156), this may not be the best choice of forms. Following the taxonomy of the family,Delphinese should probably be reserved for thelanguage of the family Delphinidae. This couldbe considered the major language group similarto the designation Romance language includesmany subsidiary languages of Western Europe. Tursiopese would be the language of the Genus Tursiops, and Truncatese would be the language ofthe bottlenose dolphin. Below that would come the dialects described by geographic location. Basedon this notation, Markov & Ostrovskaya, Hickey and Santos were studying Truncatese but Taruskiwas studying Melaenese. If it can be shown that all members of Tursiops can understand each other,the proposed Truncatese language can be demoted to the status of a dialect.

Second, most of the above authors have not provided a statistical study of the frequency of occurrenceof each sound. Only Markov & Ostrovskaya have collected enough sound samples to provide areasonable frequency of occurrence table. Only with such a table is it possible to begin attempting todecode the meaning of these sounds or groups of sounds.

Hardy Jones discussed the findings of Ken Balcom concerning the Orca of Vancouver Island74. Balcom has described three distinct dialects for the three pods of Orcas that he has been tracking formany years.

Figure 1.1.3-9 T. Truncatus waveforms identified byHickey. From Hickey, 2005.

5 September 200949

75Fripp, D. (1999)

76Watwood, S. (2003) Whistle use and whistle sharing by allied male bottlenose dolphins, Tursiopstruncatus. Ph D. Thesis MIT/WHOI 2003-14

1.1.3.3.5 Truncatus waveforms from Djurpark, Sweden

Fripp has presented a thesis study of the bottlenose dolphins in Kolmardens Djurpark, Sweden75. The goal was to explore infantile vocal development. The work focused more on social and behavioralaspects than it did on presenting waveforms. The recorded bandwidth was from 2 kHz up to “around30 kHz” using a PAL/VHS recorder. The spectrographic conversion equipment was limited to 32kHz. All of the waveforms found in this thesis were below 20 kHz and most were below 15 kHz. Time warping was used in this study to compare whistles having different time durations. Using anevent sampling approach, over 20,000 whistles were recorded.

Interesting information on the predominance of bubble-stream whistles among calves is provided.

1.1.3.3.6 Truncatus waveforms from Sarasota Bay, Florida

Watwood has recently completed a large study of paired dolphins in Sarasota Bay76. She identified125 distinctive whistles within the frequency range between 3 kHz and 20 kHz. She identified alarge group of these as associated with signature whistles. Figure 3.1 shows a selection of thesesignature whistles. She also identified the whistles most commonly shared among pairs of males. Noeffort was made to associate these whistles with any fricatives within the same frequency band thatmight lead to an understanding of more complex morphemes. Many repetitions of the same whistlewere recorded (with or without brief pauses between them). No harmonic content was presented. Noanalysis of harmonic contents was provided. A combination of human and automated comparisonswere made between the different whistles. The duration of the whistles was not a majorconsideration in this study. Figure 2.1 does provide some time information. Individual whistlestended to be 300 to 500 ms long. 1.1.3.3.7 Characteristics of tonal vs Stressed languages

The majority of human languages are divided into two large groups, the tonal and the stressedlanguages. While most Indo-European languages are stressed, Chinese, Norwegian, Swedish andmany others are tonal in character. These terms apply to individual words within the language. While many languages use variations in tone within a sentence to describe emotion, emphasis andcontrast, the use of such variations within a word indicate differences in lexical meaning.

In general, both tonal and stressed languages mature into a form containing both consonants andvowels. While languages consisting of only vowels or only consonants are known, they are rare andgenerally of limited scope. The written forms of both early Hebrew (Aramic) and Arabic began asconsonant only languages.

The spectrograms of Dolphinia (or Truncatese) strongly suggest it contains a component analogous toa tonal language. On the other hand, the spectrograms also show the dolphin makes many staccatosounds that are suggestive of a stressed language. The consequences of these situations will beaddressed more completely in a separate appendix.

5 September 200950

77Katamba, F. (1989) An Introduction to Phonology. London: Logman Group UK Limited pg 53

78Catford, J. (1988) A Practical Introduction to Phonetics. Oxford: Clarendon Press pg 183

79Deutsch, W. & Fodermayr, F. (2004) Visualization of Multi-Part Music (Acoustics and Perception). Http://www.kfs.oeaw.ac.at/fsf/mus/Poly1.htm

80Richards, D. (1986) Dolphin vocal mimicry and vocal object labeling In Schusterman, R. Thomas, J. &Wood, F. eds. Dolphin Cognition and Behavior: A Comparative Approach. Hillsdale, NJ: LawrenceErlbaum Associates Chapter 13

Amy Stafford has provided a brief tutorial on tonal languages at the website, http://www.mnsu.edu/emuseum/cultural/language/tonal.html . It includes the standardized marksfor indicating tonality.

Table 1 - Common Tone Features77 high [ / ] mid [ - ] low [ \ ] rising [ _/ ] (note underline)falling [ -\ ] fall-rise [ \/ ]

The use of these marks and the results of using them is shown below for one dialect of Chinese.

Table 3 -Mandarin Tone Use78 Word Intonation Meaning ba [/] to uproot ba [--] eight ba [\/] to hold ba [\] a harrow Stafford concludes with;

“As has been shown, there is tremendous diversity in the way that different languages around theworld are spoken. Using the same speech features, they are each able to create their own unique wayof communicating.”

Deutsch & Fodermayr have provided a selection of spectrograms of singing and chanting in variouslanguages around the world79.

This level of variation with geographic location suggests a similar problem with Dolphinlanguages/dialects.

1.1.3.3.8 Mimicry by dolphins

Richards et al. described the wide range of waveforms that the bottlenose dolphin can mimic in the200 to 30 kHz range when trained80. The range of waveforms goes well beyond the class ofwaveforms required for effective echolocation. Figure 1.1.3-10 illustrates this capability. The

5 September 200951

81McCowan, B. & Reiss, D. (2001) The fallacy of ‘signature whistles’ in bottlenose dolphins: a comparativeperspective of ‘signature information’ in animal vocalizations Anim Behav vol 62, pp 1151-1162

82Barton, R. (2006) Animal communication: Do dolphins have names? Cur Biol vol 16(15), pp R598-R599

83McCowan, B. Hanser, S. & Doyle, L. (1999) Quantitative tools for comparing animal communicationsystems: information theory applied to bottlenose dolphin whistle repertoires Anim Behav vol 57, pp 409-419

animal did have difficulty interpreting and reproducing a square wave in frame ( C). It did better inframe (I). The animal shifted the frequency of its response in frames (G) and (H). The ability tomimic a variety of signals from another dolphin goes a long way in supporting the view that dolphinscarry out communications using a language learned from their compatriots. While attempts to teacha syntax to the dolphin have not been as fruitful, it is not obvious that a language must have asyntax. Simple pattern matching may be adequate in a language of limited scope.

The dolphin began to mimic the waveformwithin a fraction of a second, usually aroundone-third of a second. It is a rare human whocan mimic another person with a delay of lessthan one second.

1.1.3.3.9 The concept of anindividual dolphin adopting apersonal noun, limited to asignature whistle

A vibrant debate has been going on since theyear 2000 concerning whether the sounds of agiven dolphin contain a signature whistleidentifying that individual. Some argue thatthe potential signature whistle may apply to agroup but not an individual. Others consider apersonal designation unnecessary within thesocial contexts they have studied. A recentpaper has considered a hierarchal generalsignature whistle for the group with a possiblesuffix for the individual81. That paperdisparaged any such whistle and took the viewthat the lexicon of the dolphin was much tobroad to allow isolation of a single signaturewhistle until more was known about thegrammar and semantics of the dolphin. Bartonhas provided a very recent contribution to thesubject and included all of the pertinentreferences82.

A second scientific investigatory path wasintroduced by McCowan, Hanser & Doyle in 199983. It began the study of the sounds of dolphins fromthe perspective of information theory. While the laboratory portions of these studies have also been

Figure 1.1.3-10 Spectrograms showing the ability of thedolphin to mimic a synthesized tone. The synthesizedsignals are to the left of the arrows. The responsestypically begin before the synthesized signal is completed,after about one-third second following the start. FromRichards et al., 1984

5 September 200952

limited to less than optimum frequencies and only whistles, the data clearly shows two majorfeatures. It shows the breadth of the possible whistle signals is much larger than addressed in theabove endeavors. It also shows the potential for non-whistles signals within a larger repertoire.

In the context of this work, all of the discussions related to signature whistles has been based only onthe whistles (vowel portions) of the dolphin auditory repertoire. Most of the records of these whistleshave been truncated to the region below 20,000 Hz by using recording equipment designed for humansound recording. Since the non-echolocation portion of the dolphin auditory spectrum extends up toat least 30,000 Hz, the data recorded to date is analogous to early human telephone communicationswhere the passband of the system was 2,500 Hz or less (as opposed to modern 3,000 Hz telephonechannels). Such telephone systems were adequate for message transmission but not for sourcerecognition. This analogy would suggest wider band recordings may show that dolphins canrecognize each other by the timbre of their voices without the need of a personal noun. Informationtheory suggests the auditory and cognitive capability of the dolphin can support a much broaderlanguage than that based only on whistle sounds. It would be startling if this additional capabilitywas not used. Physiology suggests the dolphin brain operates using a “frame time” on the order of30-50 ms. It would also be quite startling from the perspective of information theory if a dolphinshould use a minimum duration phoneme of 300-500 ms when it has a capability to interpretauditory symbols at a rate ten times higher. Kasselwitz is now recording dolphin sounds up to atleast 80,000 Hz on a relatively continuous basis. These recordings show many harmonicrelationships and nuances not recognizable from lower frequency recordings. They also show a widerange of staccato sounds intimately intermingled with the occasional whistle. Both the staccato(consonant) and whistle (vowel) sounds show structure compatible with a 30-50 ms phonemeduration.

1.2 The detailed description of the echolocation and communicationssystem of the dolphin

1.2.1 The acoustic energy generating capability of the dolphin

The dolphin family can generate at least three distinct signaling waveforms. However, the source ofthe sounds associated with these classes of waveforms is not completely decided. <The UHF energy appears to be generated by a mechanical percussion associated with the phoniclips.<The complex vocalizations at low frequency (LF), including the mimicry noted above, may originatein the larynx (as a broad band noise like source) with energy selection by the structures of the nasalpassages. This procedure would mimic communications oriented phonation in other chordates.<The swept continuous tones associated with the HF signals may be generated in the larynx or bythe phonic lips of the nares. This question remains open because of the scarcity of available data onthis subject.

When generating acoustic energy for echolocation purposes, the energy efficiency associated with theconversion of DC energy to AC energy is obviously a major factor in maximum range. However, in anoverall echolocation system, the usual quality criteria is energy efficiency per bit of informationretrieved. This latter criteria is likely to play a role in separately defining the performance of the HF(swept tonal) and the UHF (pulse) systems.

The water pressure acting on the breath-holding dolphin plays a major role in determining thepneumatic performance of the dolphin’s system at significant depths. This factor must be considered

5 September 200953

when evaluating the dolphin’s overall acoustic generation capability.

1.2.1.1 Low frequency sound generation in dolphins RESERVED

For now, the LF sound generation capability of the dolphin will be considered as analogous to that ofother chordates and involve the larynx.

1.2.1.2 Ultra high frequency sound generation in dolphins

The generation of pulse-type UHF sound has been observed most closely in dolphins. However, theintrinsic methods of sound generation and the characteristics of the sound produced were obscureuntil the last decade. It is now clear that the UHF energy is generated by the phonic lips. As notedbriefly in Section 1.1.2, Aroyan has suggested the UHF signals in dolphin are generated in analogyto the LF sounds of other chordates. He suggested a broadband noise source followed by a frequencyselective filter followed by a beam forming radiator. This arrangement is conceptually veryinefficient energy-wise since much of the energy from the source is suppressed or ignored in thephonic formation process. An alternate description of the process might be conceptually moresatisfying. Replace the broadband noise source (a random broadband spectrum) with an impulsegenerator (a correlated broadband spectrum).

An impulse source can be described as a percussion driven source. A percussion source could be usedto excite a resonant structure very efficiently (as a clapper excites a bell). This approach would notuse the pneumatic passages of the nares as a resonator. Instead, the pneumatic pressure differenceavailable would be used to actuate the percussion element. The percussion element would excite atissue structure acting as a resonator. The energy could then be transferred to the radiatorefficiently since the radiator would also be tissue based.

The following is offered as a hypothesis. It appears to be compatible with the physical conditionspresent in the dolphin.

If the contact points of the phonic lips are calcified and supported by a liquid crystallinesupport structure connected directly to the melon, the slamming together of the phonic lipscould generate a significant resonance in the liquid crystalline support structure. The precisefrequency of this energy would be determined by the physical dimensions of the containersupporting the liquid crystalline material. In the absence of any structure attached to thismaterial and forming the melon, liquid crystalline materials are known to exhibit very highQ’s. They will oscillate for long periods of time with negligible attenuation.

Under the above hypothesis, the rise time of the generated UHF energy would be controlled bythe physical characteristics of the percussion element and the liquid crystalline material. Thiscombination is likely to form a second order physical system and the energy as a function oftime is likely to exhibit an S-shaped leading edge. If the acoustic coupling between theresonator and the melon is considered tight, the energy of the liquid crystalline material will berapidly transferred to the melon for projection into the water. This will result in a rapiddecline in the energy stored in the resonator. As a result, the energy transferred to the melonand then the ocean will exhibit an S-shaped leading edge followed by a rapidly falling trailingedge. The composite shape of the waveform expected from such action can be expected to looklike a Gabor-shaped resonance in the absence of more detailed information.

As generally accepted, a Gabor shaped pulse is of maximum value in the operation of a background

5 September 200954

84Au, W. (2004) Op. Cit.

noise limited monopulse echolocation system. It can be used equally efficiently in a multi-pulse trainsystem where additional information may be extracted from the pattern of the echoes received.

1.2.1.3 High frequency sound generation in dolphins

It is difficult to discuss the HF sound generation capability of the dolphin based on the very limitedinformation available concerning the detailed morphology employed. It could be generated usingfrequency selection associated with a wideband pseudo-noise source as assumed for the LF situation. Alternately, it could be generated as a mechanical resonance as proposed for the UHF system. Another possibility is that it is generated as a conventional pneumatically-based whistle. Themultiple large and dynamic air sacs in the head of this breath-holder appears particularly adapted tothis role.

Neither the tailored noise source used at LF or the percussion-type resonator proposed for use atUHF are well suited to the generation of swept continuous tones at high power levels. The movementof air in significant volume passed a resonator of the type found in pipe-organs or brass musicalinstruments appears most appropriate. Of course, the optimum situation is where the energy isconcentrated primarily in the tissue medium of the head rather than in the gaseous volumes. Suchenergy can be transferred from the source to the water most efficiently.

The generation of a swept tone by the larynx appears feasible if air can be transferred back and forthbetween the sacs of the nares and the lungs. Alternately, the generation of a swept tone by thephonic lips appears feasible if air can be transferred back and forth between the pre-maxillary sacsand the sacs closer to the blowhole and the phonic lips are compatible with the generation of acontinuous tone over a period of a few seconds. The latter approach would require the phonic lips beheld rigidly at a very close spacing while the liquid crystalline or other supporting materials wereallowed to resonate.

A question arises as to whether the vocal chords of the larynx could be suitable for supporting a vocalrange from 200 Hz up to 30,000 Hz, particularly at high power levels and efficiency. An alternatechoice would be to use the phonic lips hypothesized above in a fixed configuration in conjunction withthe continuous transfer of air. The frequency of the oscillation could by varied by varying thephysical dimensions of the liquid crystal also hypothesized above.

Current exploratory research appears to be leaning toward the generation of high intensity HFsound by the phonic lips. The above discussion suggests a common acousto-mechanical mechanismcapable of supporting both HF and UHF sound generation.

1.2.2 The beam forming capability of the dolphin

The technical literature has made steady progress in defining the beam forming capability of both thetransmitting and receiving systems in the dolphin. The progress has been so great that thedescriptive literature prior to the year 2000 must be largely discounted. The system is considerablymore complex, and of much higher performance, than that referred to conceptually by Au in a poorlytimed paper84 (2004). Both the transmitting and receiving systems incorporate a variety of functionalelements and resulting capabilities not considered by Au. The expansion of these capabilities

5 September 200955

85Johnson, S. (1986) Dolphin audition and echolocation capacities In Schusterman, R. Thomas, J. & WoodF. eds. Dolphin cognition and behavior: a comparative approach. Hillsdale, NJ: Erlbaum Assoc pp 115-136

explains why the signal processing and signal manipulation capabilities of the dolphin neural systemhave been expanded so significantly. These capabilities have resulted in a brain approximately thesame size as the human brain (in an animal about three times larger in mass) but containing a verydifferent allocation of resources (Section 1.4.4).

Bullock & Gurevich reference Ayrapet’yants, et. al. (1969) in describing “some of the features of thesystem in solving near-threshold discrimination; shorter echolocating pusles (30-150 μsec), changesin the pulse envelope, in the frequency composition, in the number of principal sinusoids in the pulse,and in the modulating sinusoids in the pulse, and scanning head movements which began 7-8 maway from the target. They emphasize that the directionality of the beam can be varied widely andits direction can be rapidly changed without head movement.” This is the only strong statementfound that would suggest the beam-forming system is as dynamic as modern “track while scanning”radar sets.

The following sections will show the click generating capability involves two distinct and parallelsound generators. The literature reports these generators can be used in synchronism orsequentially with different tone frequencies. Some authors claim these two UHF generators can beused simultaneously with a HF generator.

1.2.2.1 High frequency, wide beam, echolocation in dolphins

The operating characteristics of the high frequency (HF) sonar, using a swept tone within 6 to 30kHz, of the dolphin has not been well characterized to date. In fact, it is not agreed whether the HFsignals are generated in the larynx or in the phonic lips of the nares. With this ambiguity, it isdifficult to define the beam forming capability of the HF system. Based on the lower frequency, andlonger wavelength, of the energy relative to the size of the dolphin, the beam forming capability canbe expected to be considerably poorer than for the UHF capability.

1.2.2.1.1 The anatomy of HF echolocation in dolphins RESERVED

1.2.2.1.2 The schematic of HF echolocation in dolphins RESERVED

1.2.2.2 Ultra high frequency, narrow beam, echolocation in dolphins

1.2.2.2.1 The anatomy of UHF echolocation in dolphins

Figure 1.2.2-1 shows the general layout of the echolocation system of the Atlantic bottlenosedolphin, Tursiops truncatus85.

5 September 200956

The system has evolved nearly beyond recognition compared to that of other mammals. Theanatomical feature commonly called the outer ear is no longer used for hearing but may perform afunction akin to a pitot tube in aviation (measuring the relative pressure of the externalenvironment. The energy collected in support of hearing is received through a pair of gradient indexLuneburg lenses aligned along the lower jaw. Similarly, the sound source is not in the larynx but inadditional structures located between the larynx and the blowhole. Note the external opening of thistube is closed during click generation (and during lower frequency underwater vocalization).

The generated energy is transferred to the water in a narrow beam through a system similar inconcept to the lens system of the mammalian eye. This system forms the large structure above themouth of the dolphin (the melon and the tissue covering the melon). The melon contains a largeasymmetrical gradient index lens (Au, p. 88) separated from the exterior environment by a curvedsurface reminiscent of the cornea in function. This outer surface forms a lens of significant refractingpower. This lens also introduces significant bending of the beam in order to create a far-field beamcentered directly ahead and about 5 degrees above the centerline of the animal. The formation of thetransmitted beam is highly frequency dependent. The pattern can be widened by more than 2:1 by

Figure 1.2.2-1 The echolocation geometry of the bottle-nosed dolphin. The system employs lenses instead of hornsto create the radiative patterns used for both echo generation and echo reception. The transmission beam isapproximately 10 degrees at the 3 dB point in both planes at 120 kHz. The nominal gain for these angles is 26 dB. The reception beam is slightly broader in both planes at 120 kHz. Modified from Johnson, 1986.

5 September 200957

86Cranford, T. (1988) The anatomy of acoustic structures in the spinner dolphin forehead as shown by X-raycomputed tomography and computer graphics In Nachtigall, P. & Moore, P. eds. Animal Sonar: Processesand Performances. NY: Plenum Press pp 67-77

lowering the frequency of the signal to 30-60 kHz. The equivalent aperture of the transmittingsystem is a single circular plate of 2.0 to 5.0-cm diameter depending on species.

Only limited data is available on the operation of the lens system. Norris & Harvey have provideddata on the velocity of sound through small sections of the melon by dicing it into 1.8 x 1.8 cm x 9.5cm pieces. Since the wavelengths of interest are as short as one centimeter and field patterncalculations require information at no larger than one-quarter wavelength intervals, this data canonly be taken as indicative of the acoustic properties of the lens. They did record velocities thatwould suggest acoustic indices of refraction on the order of 1227/1500 = 0.818 to 1761/1500 = 1.174after correction for normal body temperature and taking the speed of sound as 1500 meters/sec understandard ocean water conditions. This wide range of indices, above and below 1.00, would suggest alens of considerable refractive power, comparable to the lenses of the visual eye (nominal indexranges of 1.45 in acoustics and 1.33 in vision).

The symmetry of the air sac system (and the asymmetry of the connections of the melon to the nares)would suggest a left and right nasal plug. In addition, the imagery of Cranford86, shown in Au (p. 89)suggests an even more complex interface between each half of the melon and the trachea. Thesefeatures suggest the transmitting system of the dolphin may employ two phase-related sources in thegeneration of the overall radiative sound pattern ( a more complex arrangement than characterizedby Au on pages 90-91). A pair of phase coherent lateral sources would make horizontal beamformation in the high frequency regime considerably simpler and might introduce additional featureswell known in the radar and sonar communities.

The receiving system employs a similar acoustic system. The initial components consist of theexterior pliable material directly below each side of the mouth and an interior fatty materialcontrolling the propagation of energy into the region of the acoustic bulla. The energy is delivereddirectly to a window in the vestibule. This energy traveling as a longitudinal wave is converted into asurface acoustic wave on the surface of the tectorial membrane of the cochlea within the vestibule. As in other marine vertebrates, the dolphin does not require the impedance transforming function ofthe middle ear. The receiving system can be modeled using an effective collection area of 8.1 cm2 foreach ear at a separation of 12 cm.

Au shows the temporal pulse signatures of a series of dolphins on pages 134-135. A typical pulseconsists of a Gaussian (or Gaussian-like) pulse of 8-10 cycles followed by one or two pulses of about 4-5 cycles at a lower intensity. The pulse duration is commonly within the typical 30 ms integrationtime found in many neurological systems. This is a different integration time than used by Au onpage 158. That integration time might be more appropriately called a correlation time in the contextof this work.

The reflection of a typical signal, as defined above, from a target of variable internal density can bequite complex and the dolphin appears to be able to take advantage of this complexity to identifyobjects successfully at signal-to-noise ratios greater than 10 dB.

Figure 1.2.2-2 shows a more detailed lateral view of the dolphin head with its echolocation elements

5 September 200958

87Norris, K. (1968) The evolution of acoustic mechanisms in odontocete cetaceans In Drake, E. ed.Evolution and Environment. New Haven: Yale University Press pp 297-324

highlighted87. It is currently believed that the sound generated by the system is independent of thelarynx. The sound appears to be generated by the presence of the nasal plug formed by an extensionof the melon. This plug vibrates as the result of air being forced passed the plug within the tubularsac. The air is squeezed out of one or more of the various sacs found in this region and received byone of the other sacs. It is believed the air is re-circulated to allow indefinite operation of the systemunder water. The energy generated at the nasal plug is focused initially by the shape and thegradient index of the material forming the melon. The formation of the beam is completed by thelens formed by the outer skin/water interface. The reflected energy is captured by the mandibularwindow of the lower jaw. This window also acts as a lens in conjunction with the water-to-tissueinterface. The captured energy is then focused further and transferred to the area of thetympanoperiotic bone by the fatty channel.

Figure 1.2.2-2 The general layout of the echolocation system of the dolphin. The fatty cavity forward of themandibular window is probably not related to hearing. Modified from Norris, 1968.

5 September 200959

88Lemerande, T. (2002) Transmitting beam patterns of the Atlantic Bottlenose Dolphin (tursiops truncatus):Investigations in the existence and use of High frequency components found in Echolocation signals. NavalPostgraduate School, Monterey, CA 93943-5000

1.2.2.2.2 The potential for rapid beam steering

Several authors have described high speed sweeping of the UHF beam of the bottlenose dolphin. Such sweeping could be achieved by several methods. Figure 1.2.2-3 shows in caricature the abilityto change the beam direction based on the asymmetry in the locations of the dual sources of thebottlenose dolphin (See Section 1.1.2.3.3). Without detailed knowledge of the location of the sourcesand the feed structure between the sources and the melon, this beam steering mechanism cannot bedemonstrated or quantified. An equally possible method of beam steering involves muscular changein the shape of the melon or the feed structure. A third possible method involves changing thelocation and content of the air sacs (acting as mirrors) within the forehead.

Bullock & Gurevich (p. 74) review some of these possibilities and the remarkably high rate of beamsteering reported without documentation by Reznikov. Reznikov claimed the beam could be rapidlyrotated at a rate of 3.5 x 105 degrees/sec independent of any head motion. Such a high rate might begenerated by interference between the two sources operating at slightly different frequencies. Suchhigh rates due to muscular action seem unlikely unless the melon is operated in a multi-lens foldedtelescope configuration.

1.2.2.3 Potential operation in the 150-390 kHz

Lemerande has provided some very interesting field research results at frequencies above thoseexplored before88. He makes the following assertion. “EFSDLs at 390 kHz clearly show the dolphinhas the capabilities to actively use frequencies almost three times higher than previouslydocumented Tursiops Truncatus sound emissions to echolocate a target.” This assertion is madeentirely on the basis of his recording energy emitted by the dolphin as measured at ranges of a few

Figure 1.2.2-3 Potential beam direction variation in the bottlenose dolphin based on two distinct source locations. The sources may also have a lateral displacement that may also affect the beam direction. Potential changes in beamdirection/form due to muscle activity are not shown. From Goodson & Klinowska, 1989.

5 September 200960

89Murray, S. Mercado, E. & Roitblat, H. (1998) Characterizing the graded structure of false killer whale(Pseudorca crassidens) vocalizations J Acoust Soc Am vol 104(3), pp 1679-1688

90Novick, A. & Vaisyns, J. (1964) Echolocation of flying insects by the bat, Chilonycteris parnellii. BiolBull vol 127, pp 478-488

91Novick, A. (1964) Echolocation of flying insects by the bat, Chilonycteris psilotis. Biol Bull vol 128, pp297-314

meters. He did not consider the two-way attenuation of the signals or the operation of the dolphinsauditory system.

1.2.3 The range measuring capability of the dolphin

Murray, Mercado & Roitblat provided a simple computational model describing the rangingcapability of the false killer whale89. Although they do not introduce a feedback loop within theauditory system, Murray’s team offered a number of useful observations concerning the soundsemanating from the false killer whale. They noted the ability of the animal to generate bothcontinuous tones and simultaneous pulses within the frequency range below 22 kHz. They also notedthat when the animal generates a series of pulses with spacing of less than 5 ms, humans perceivethis signal as a continuous tonal sound.

A key feature of ranging by marine mammals appears to be the overlap between the pulsetransmitted and the received signal as an echo from the previous pulse within the inferior colliculus. This time coincidence allows the animal to adjust the pulse-to-pulse interval to center the nexttransmitted pulse relative to the echo. This adjustment can be made on a pulse-to-pulse basis. As aresult, the pulse-to-pulse interval of the pulse generating system (averaged over several pulses) isdirectly proportional to the distance of the target from the animal. This signal is produced by asimple servomechanism without any counter circuits or complicated correlation calculations.

Not many investigators have focused on the relative time between the receipt of a ranging echo andthe time of generation of the next output pulse. To measure this feature requires specialized testequipment and a detailed knowledge of the physiology of the animal involved.

Novick has analyzed the pulse overlap characteristic in the ranging servomechanism of the bat,Chilonycteris parnellii90 and Chilonycteris psilotis91. All phases of the search, pursuit and capture(terminal) sequence are discussed.

The multiple air chambers in the head of the dolphin and the isolation of the cochlea within bullaisolated from the skull are very effective in isolating the high power transmitting capability of thedolphin from the high sensitivity of the receiving system. This isolation is further supported by thelocation of the beam-forming melon above the aural cavities and the two beam-forming receivingstructures below the aural cavities.

1.2.3.x The related range measuring capability of the bats

Although the repertoire of techniques used by the bats differ in some aspects from those used byCetaceans, a comparison is appropriate.. In some cases, such a comparison offers additional insightinto the physiology and capability of Cetacea.

5 September 200961

The very detailed data provided by the Novick team is typified by Figure 1.2.3-1. The portion of thefigure to the right of the arrow relates to the acquisition mode of the bat. The time from the arrowdown to 125 ms represents the pursuit phase. The terminal phase begins at 125 ms. These phasedesignations differ slightly from Novick as discussed below. A problem appears in the caption of mostof their figures. They provide the expression “One msec = 1.73 mm.” This should be “ bat movementduring 1 msec = 1.73 mm.” or “bat flight speed = 1.73 mm/ms” as given in their Table 1. The round-trip travel time for an acoustic signal is 100 times higher at 170 mm/ms. Based on their assumptionof a constant bat velocity during their recording sequences, an auxiliary scale has been added to theleft of the figure. This scale suggests that these bats, like other species, depend on the sounds madeby their targets for locating food at ranges exceeding one meter. Thus, the bat’s search mode usespassive sonar while the acquisition, pursuit and terminal phases of the predation mode uses activesonar. This situation suggests this bat uses the characteristic sounds of its prey to classify its targetsduring the search mode. The data clearly shows this bat uses a variety of swept frequency pulsesduring the initial acquisition, the pursuit and the terminal phases of an encounter.

Novick speculates that the bat may use a servomechanism based on pulse overlap to determine therange to the target on page 307. He notes this technique is independent of the speed of sound anddoes not require the ability to discriminate short intervals of time. This bat used a fundamental toneof 21 kHz plus the 2nd, 3rd and 4th harmonics. “Either the second or, more often, the third harmonic isat the highest amplitude, especially during the terminal phase (Fig. 1).” Pulse to pulse frequencystability appears to be better than 1 kHz. Intra-pulse, the tone frequency drops from 21 to 17 kHz ina generally linear sweep.

5 September 200962

Roverud & Grinnell have provided a pair of papers discussing the detection and ranging capabilities

Figure 1.2.3-1 An example of a fruit fly pursuit by C. Psilotis. Pulse duration and interpulse interval are plottedagainst time before the end of the pursuit. Assumed bat movement 1 ms = 1.73 mm. The projected target range iscalculated from this value. The arrow indicates the first calculated pulse-echo overlap. Note; time reads from rightto left (larger numbers represent earlier events). From Novick, 1965.

5 September 200963

92Roverud, R. & Grinnell, A. (1985) Discrimination performance and echolocation sighal integrationrequirements for traget detection and distance determination in the CF/FM bat, Noctilio albiventris. J ComPhysiol A vol 156, pp 447-456

93Roverud, R. & Grinnell, A. (1985) Echolocation sound features processed to provide distance informationin the CF/FM bat, Noctilio albiventris. J Com Physiol A vol 156, pp 457-469

94Simmons, J. Ferragamo, M. Moss, C. et al. (1990) Discrimination of jittered sonar echoes by theecholocating bat, Eptesicus fuscus: The shape of target images in echolocation J Comp Physiol A vol 167,pp 589-616

of the bat, Noctilio albiventris92,93. This bat is a member of the order Microchiropteran. Simmons etal. have provided a paper containing much data on a bat of the order Chiropteran94. However, theiranalyses are based on the use of techniques adopted from conventional man-made acoustic rangingequipment. As noted in the introduction to this appendix, bats (along with other mammals) are notcapable of computing autocorrelation functions using transcendental calculations.

Using the gated time window technique described by Roverud & Grinnell instead of the differentialtime measurements of Simmons et al. leads to a physiologically feasible echolocation system.

1.3 The overall neural physiology of echolocation in dolphins

The transmitting and receiving functions associated with the dolphin offer a very wide range ofpotential signaling capabilities. It is important to consider these potentials in order to understandthe operation of the neural system.

It has been shown (Fulton, 2004, 2005 & 2006) that the neural system is not capable of performingtranscendental mathematical calculations (except for a natural logarithmic conversion) and containsno resonant circuits. The neural system is entirely temporal-based. The neural system is notcapable of performing calculation involving sines and cosines and it is not able to perform Fouriertransformations in the frequency domain.

It has also been shown that the neural system is place (as opposed to frequency) oriented. In bothvision and hearing, the signals generated at specific locations within the retina and cochlea areprocessed by the remainder of the neural system. In this sense the systems are either retinotopic orcochleotopic rather than directly related to the parameters of the environment ( such as tonotopic). This is important because, merely changing the curvature of the cochlea can change the frequencyrange to which the auditory system is sensitive without making any changes in the remainder of theneural system.

It should be appreciated that most of the following material in this section is based on inferences,based largely on comparative anatomy, that require future laboratory demonstration.

The neural system operates as a “pipeline” signal processor. Once data has been acquired at thecochlea, the signals are passed along multiple parallel channels within the system. There is norequirement that the processing within these channels be completed before the next data acquisitioncycle begins.

It appears that most bottlenose dolphins do not emit a new acoustic signal until 19 to 30 ms after thereflection associated with the previous emission has been received at the outer ear. This is not truein the case of a beluga whale, Delphinaterus leucas, as shown by Turl & Penner, 1989. Figure 1.3.1-

5 September 200964

1, from Turl & Penner, shows one of the echolocation signal patterns used by a beluga whale wheninvestigating a target at a range of 100 meters. The pattern shows the beluga sends out a series ofclosely spaced pulses and then pauses. This pause allows the last pulse to reach the target andreturn before any other pulse is emitted. This method of operation allows the beluga to determinethe unambiguous range to the target, using the last pulse in each group, while achieving a highersignal to noise ratio with respect to the target by using multiple pulses where the target range couldbe interpreted ambiguously. This method should work best in open ocean situations wherereverberation is not of major concern. The beluga employs two other pulse patterns that may bemore useful in high background situations. Note the pulse spacing in the lower frame. The intervalbetween pulses is near the fusion interval of the human visual system. It is likely that pulses aresent at intervals of near 30 ms to insure a perception of continuity between the individual perceivedacoustic “images.”

5 September 200965

1.3.1 Intrinsic signal generating capability of the dolphin

The transmitting and receiving functions associated with the dolphin offer a very wide range ofpotential signaling capabilities. It is important to consider these potentials in order to understandthe operation of the neural system.

The signals that can be generated by the phonation system include those of both the larynx and thetwo mechanisms associated with the spiracular cavity. The basic forms of these signals include:1. A complex broad band low frequency (LF) signal, 200 Hz to 15,000 kHz, not unlike that generated

Figure 1.3.1-1 A beluga whale click interval pattern for a target at 100 meters. Top; click interval versus clickinterval number. Bottom; amplitude versus time profile. From Turl & Penner, 1989.

5 September 200966

95Moore, P. & Pawloski, D. (1989) Investigations on the control of echolocation pulses in the dolphin,Tursiops truncatus In Thomas, J. & Kastelein, R. Op. Cit pp 305-316

by other mammals of its size.

2. A swept narrow band signal operating within the HF range of 4,000 to 30,000 Hz.

3. A band limited impulse type signal occurring in the UHF range of 30 kHz to 150 kHz.

The literature is clear that at least two and probably three sources of acoustic energy can operatesimultaneously in the dolphin. There is no question that at least two impulse type signals can begenerated simultaneously as well as sequentially.

The complexity of the emitted signals makes their analysis, even their display prior to analysis byhumans, awkward as well as difficult. The dynamic range of the emitted patterns as a function oftime is difficult to capture in a display that also captures the long term signal characteristics. Figure 1.3.1-2 is a poor copy of a waterfall display that illustrates the problem95. Only energiesabove 205 dB were displayed in this presentation and the animal was not allowed to approach thetarget. Moore & Pawloski also reported on their ability to train the dolphin to modulate its peakpower output according to a pattern if not on command.

Frame A shows an initial few pulses that are unimodal. The sequence changes to a bimodalstructure around click 5-10. This pattern is maintained until about pulse 85 when it returns to amonomodal form.

1.3.2 Potential signals received by the dolphin from its environment

Figure 1.3.1-2 Waterfall displays of the energy distribution in emitted clicks with a 205 dB threshold. An alternatedisplay would present time on the horizontal axis. A similar display can present time on the vertical axis instead ofclick number. Each click is plotted vertically as a series of horizontal spectra. Three shades of gray represent the20, 10, and 3 dB energy bandwidths. A; a click train of 101 clicks emitted during a search with no intentional targetpresent. B; 29 clicks emitted during a target present trial. The animal left the station to report target acquisition. From Moore & Pawloski, 1989.

5 September 200967

The dolphin is exposed to and can be expected to process both passive signals generated by itsenvironment and active signals generated by itself. The signals generated by the environmentinclude potential communications signals from other members of its species. Because of the socialcharacter of the dolphins, it can be expected that the echolocation signals generated by one dolphinwill be received by other dolphins and must be rejected as extraneous for their purposes.

This section will develop the characteristics of the potential signals used in dolphin echolocation. Itwill be seen that both time difference measurement (based on pulse transmission techniques) andspectrographic measurement techniques (involving frequency comparison and correlation processing)offer considerable utility to the dolphin.

The ability of the bottlenose dolphin to vary the character of its emitted signals dynamically makesidentifying the features in a given received signal difficult. The received signal can vary from pulseto pulse, and quite possibly within the echo pulse due to frequency and amplitude changes within theemitted pulse generating the response.

1.3.2.1 Character of exogenous signals

The exogenous signal environment could be of interest to the dolphin for a variety of reasons. Thereception of passive signals using binaural receiving techniques (even in the presence of interference)is well known from the experience of humans. However, the detailed analysis of this area is beyondthe scope of this study of echolocation.

1.3.2.2 Character of the reflected endogenous signals

An active signal reflected from targets within the environment can be analyzed using a variety oftechniques involving both time delays and frequency shifts. These capabilities have been widely usedin man-made equipment using fixed frequency and swept frequency continuous tone signals as wellas single and multiple pulse signals. It is appropriate to consider if and how the dolphin couldemploy these same techniques.

While it is well known that a target can be tracked by its reflected Doppler signal generated inresponse to a continuous fixed frequency tone reflected from a “moving” target. It is less well knownthat a target can be tracked by the change in received frequency at a given time in response to thetransmission of a swept continuous tone signal.

It is also well known that a target can be tracked using the energy reflected by the target in responseto interrogation by a pulse signal of known timing. Man has developed a wide range of displaytechniques allowing the extracted information concerning the targets to be presented to the senses ofthe human equipment operator.

All of the signals described above are processed more effectively using binaural receiving techniques.

1.3.2.2.1 The reflective characteristics of fish

Textbooks frequently describe the reflection of acoustic energy from fish in analogy with thereflection from the skin of aircraft in radar applications. Reference is also frequently made to thereflections from the moving appendages of fish. In general fish (including their appendages) havedensities very similar to that of water. As a result the reflected signal, due to the impedance changeat the surface of the fish, is quite small compared to that of an airplane. The situation is analogous

5 September 200968

to the modern stealth aircraft that employs composite materials to avoid radar reflections. Whilesuch a reflection can be detected, there is a more important reflector involved. The so-called swim-bladder, an internal bladder filled with air, represents a major impedance discontinuity between thedensity of the fish and the air. As in the case of air sacs within the head of the dolphin, this swim-bladder acts as a very effective reflector of acoustic energy. The bladder plays a role analogous to theactive transponder used in modern aircraft operations.

1.3.2.2.2 The character of reflected continuous tones

The detection of targets illuminated by continuous tones involves the sensing of differences infrequency between the radiated and received signals. These differences can involve a simple timedelay due to the distance to the target or a shift in frequency due to a difference in relative velocitybetween the source and the target. Both of these changes are readily detected by a system withadequate means of remembering the characteristics of the transmitted signal at a given time. Figure 1.3.2-1 caricatures some of the features of the signals involved in this mode of operation.

Frame A of the figure illustrates a simple continuous tone sonar (top line) and potential reflectionsfrom three targets. Target T1 is a fixed target. The frequency of the reflected signal is the same asthe transmitted signal. The distance to the target can only be ascertained roughly as a multiple ofthe phase difference between the radiated and reflected signals (if within the bandwidth of thereceiving circuitry). Target T2 is closer to the emitter and moving toward it as shown by the largeramplitude and higher frequency of the received signal. Target T3 is moving away from the emitter asshown by the lower frequency of the received signal. This system has low utility because of theambiguity in the range estimation.

Frame B shows a more utilitarian swept continuous tone system. The source is shown as a dashedline. A nearby target generates a reflection as shown by the left solid line. Such a reflection mightindicate a nearby surface. The right portion of the figure shows a more interesting situation. Thereflected signal complex consists of a continuous swept tone representing a simple target at a longerrange. Associated with this signal are additional signal components associated with elements of thecomplex. These signals can be associated with other elements appearing and disappearing in thevicinity of the major target or as elements of the major target moving toward the emitter at differenttimes and elements moving away from the emitter at different times. This could be a representationof a school of fish swirling about a common average range. In that case, the reflected pattern fromthe overall complex at a given time might resemble an ellipse as shown by the dotted line. It will beseen that this pattern is not significantly different from those associated with a chord of music, a setof tones occurring at a given time. Such musical patterns are readily identified by humans.

A swept continuous tone system is capable of good range resolution. Using the data of Caldwell(1990), the sweep rate for the bottlenose dolphin is on the order of 180-250 kHz/second. By storing aneural rendition of the emitted signal and comparing it with a received reflection, an estimate of therange of a target can be made. Such an estimate can be made simultaneously for every target withinthe range of the system using correlation techniques.

Simple envelope detection does not provide significant range information in a continuous amplitudetonal system, whether using a fixed or swept tone. However, the dolphin uses a frequency dispersionsystem to detect tonal energy within each of many narrow bands to which it is sensitive. Envelopedetection of this energy can be used for ranging as well. The differential associated with the leadingedge of these reflected envelopes can be compared with the reference sample with considerable

5 September 200969

accuracy. A value of 50 microseconds is frequently found in human psychophysical and animalelectro-physiological tests. If achieved in the tonal channels of the dolphin, this accuracy wouldsuggest a range accuracy on the order of five to ten centimeters for the swept tonal system.

Frame C will be discussed in the next section.

The swept continuous tone biosonar, with the tone lasting up to a few seconds, appears admirablysuited to the needs of the dolphin. It provides a large amount of information concerning a targetcomplex in a form similar to that known to be used effectively by other animal auditory systems. Italso allows the separation of multiple targets and backgrounds based on concepts similar to thoseknown to be used by humans to separate a desired complex communications signal in the presence ofinterference.

5 September 200970

Figure 1.3.2-1 The characteristics of various reflected signals expected in biosonar applications. A; a continuousfixed frequency signal and its reflection from three targets. B; a continuous swept frequency signal and severalreflections. C; a pulse signal reflected from a target and clutter due to the background. See text.

5 September 200971

1.3.2.2.3 The character of reflected impulse signals

The basic monaural and binaural processing of reflected pulse information by animals is based on thetime of arrival of the reflected signal without regard to the phase of the high frequency content of thepulse or of the relative phase between the high frequency content of the signals received at the twoears. There is an exception to this for signals at frequencies below 600 Hz (and possibly below 1,000Hz). The early circuits of the auditory system are capable of reporting amplitude variations (andtherefore phase relationships) at frequencies below this limit. As a result, the binaural capability inhumans can be shown to use both the time of arrival and the phase difference information (for lowfrequencies) in its determination of source angular location. It would be expected that the samecapability would be available to dolphins.

Frame C of the above figure illustrates the monaural signals involved in an impulse-basedecholocation system. Following an initial high intensity pulse from the emitter, a complex ofreflected signals is typically received. These signals typically consist of a signal from an unidentifiedtarget plus signals from the local surface and bottom of the sea (generally described asreverberations) and signals from other extraneous objects in the background. The background andreverberations are frequently described as clutter. The problem of target isolation is frequentlysolved by emitting multiple impulses and performing frame-to-frame subtraction on the receivedsignals to isolate moving targets. Once a target (moving within an appropriate velocity with respectto the emitter) has been isolated, simple windowing techniques, similar to those used in man-madesystems, can be employed to suppress the clutter and emphasize the return from the target.

While it is possible to extract both position and velocity information from signals generated by thereflection of continuous trains of impulse, the process is difficult and incompatible with the basicarchitecture of biosonar. Impulse-based biosonars appear designed specifically to exploit single pulsemapping techniques.

1.3.2.2.4 Useful biosonar configurations

Based on the above discussion, it appears obvious that the dolphin would be well served by either aswept continuous tone type of sonar or a simple monopulse type of sonar. If it was able to combinethese two types, additional advantages would probably be obtained that are not obtainable from onlyone type. Based on the combination of whistle and click signals emitted by most dolphins, it appearsthis dual capability has been achieved.

Caldwell, et. al. have noted in Leatherwood & Reeves that infant dolphins do not exhibit a signaturewhistle (an orderly swept continuous tone signal). However, they develop a signature whistle wellwithin their first year. This development would be compatible with the process of appreciating theutility of the responses to such signals and memorizing (learning) the patterns received as they relateto food sources.

1.3.3 Extensions based on binaural and biphonic implementations

The mammalian echolocation system employs two techniques not found in typical man-made radarand sonar systems, the use of two separated receiving apertures (as in most lateral animals) and twoindependent transmitting sources forming a beam through a common projection system.

5 September 200972

Humans are known to use binaural hearing for passive source location and signal discrimination. Both phase differences and time of arrival differences are used. It must be assumed that thedolphins share this capability.

The use of two transmitting sources simultaneously could generate controllable interference patternsin the far field pattern of the transmitting system that would significantly change the short termreflected energy received as a function of distance from the sources. Such a system could provideadditional information about the longitudinal dimensions of any targets. Signals generated by thedual beams of a dolphin have been reported, but based on only unsophisticated instrumentation.

1.3.4 The potential neural architecture of the dolphin

The overall target location capability of the dolphin can be estimated initially by comparing itsmorphology with that of other mammals, particularly the human (extensively analyzed behaviorally),the cat (extensively analyzed physiologically) and the bats (extensively analyzed both behaviorallyand physiologically).

It is likely that the capabilities of the dolphin to analyze the non-location aspects of signals from itspassive environment are similar to that of other mammals. Such perception is focused in thediencephalon (perigeniculate nucleus and pulvinar of the thalamus) and in the temporal lobe of thecerebral cortex.

The ability of the human and other mammals to locate sources is well quantified. This ability hasbeen traced to the binaural capability of the generic auditory system residing largely in the inferiorcolliculus. This capability has been found to consist of two separate sub-capabilities. One sub-capability, extending up to about 600 Hz in humans, employs phase sensitive binaural signalprocessing in the tonal signaling portion of the system. The second sub-capability relies uponrelative time delays measured by the binaural component of the temporal signaling system. It islikely that the dolphin is able to utilize these same passive capabilities.

The ability of a system employing an impulse type illumination source could operate much like theabove passive sub-capability based on the temporal portion of the binaural system. Such a capabilityis likely to be found associated with the inferior colliculus or its homologs.

The ability of a system employing an active swept continuous tone source could operate much like theabove passive sub-capability based on the tonal portion of the binaural system. This is a capabilitynot found in mammals other than possibly in the bats. It would be expected to require significantneural signal manipulation capability. The logical source of this capability would be the laterallemniscus, an engine largely without known purpose and nearly vestigial in humans and many othermammals. The lateral lemniscus are large and highly elaborated in the dolphin.

It is proposed the information acquisition portion of the hearing system of the dolphin;< is highly developed for both communications and target location,< is capable of target location using both passive and active techniques,< uses both swept continuous tone and impulse forms of biosonar for echolocation, and< employs both binaural and biphonic techniques beyond the scope of most man-made sonars.

It is proposed the neural signal generation portion of the hearing system;< is typical of that used in other mammals,< is uniquely shielded by reflective air sacs from the high intensity acoustic generation function,

5 September 200973

96Fulton, J. (2005) Processes in Biological Vision. www.sightresearch.net

< employs dispersion (rather than resonance) techniques for signal separation within the cochlea.

It is further proposed the information extraction portion of the hearing system of the dolphin can bedescribed as consisting of three parts,<The first part is focused on intraspecies communications with activity centered in the thalamus ofthe diencephalon and the temporal lobes of the cortex (in common with other mammals).<The second part is focused on passive source location and precision pointing operations with activitycentered in the inferior colliculus (in common with other mammals) and other elements of the PES(shared only with the bats).<The third part is focused on (acoustic) target imaging and analysis (using techniques homologouswith those used in communications analysis) with activity centered in the lateral lemniscus of thebrain.

Such a biosonar system significantly exceeds the capability of current man-made systems.

1.4 The neural physiology of specific circuits used for echolocation indolphins

The block diagram of dolphin echolocation and the generic schematic of mammalian hearingdescribed in Section 1.1.2 can be expanded into detailed discussions of the individual engines of thedolphin neural system. To do this requires some background in the electrical operation of the basicneuron. It also requires recognition of the basic architecture of neural signaling. Finally, puttingmatters in perspective is easier if it is noted that each neural center dedicated to a specific taskconsists of more than four million individual neural circuits. Because of the overall complexity of theseindividual centers, they are defined as engines in this work. This name correlates very well with the engines definedin current computer processing. An engine is a major hardware portion of a computer that incorporates memory aswell as algebraic manipulation and frequently employs its own specialized software routines.

Neurons can be divided into three basic groups, the sensory neurons, the signal processing neurons,and the signal projection neurons. The sensory and signal processing neurons operate in the analogmode. Only the signal projection neurons operate in the phasic mode (and generate actionpotentials).

Within the cortex, projection neurons are used to interconnect the different feature extraction engines and logic unitsof the brain with the initial motor system command generation centers. Projection neurons typically connectremote neural engines (separations larger than two millimeters).

These concepts are explored in greater detail in Chapter 8 of “Processes in Biological Vision96.”

1.4.1 Overview of the electrolytic neuron

The electrolytic theory of the neuron presented in “Processes in Biological Hearing,” (the parent ofthis paper) has provided an entirely different context for neural operation than the previouschemistry-based hypothesis. The electrolytic context allows description of the operation of the neuralsystem at a level never addressed under the chemical hypothesis.

1.4.1.1 Basic physiological forms of neurons

5 September 200974

The fundamental neuron is a three-terminal electrical device. The principal terminals of the deviceare the dendritic terminal, the poditic (base or foot) terminal and the axon terminal. Together, thedendritic and poditic terminals are called neurites. In the morphology of a typical neuron with twoneurite arborizations, one is dendritic and the other is poditic. The neuron forms a circuit similar toa conventional transistor circuit and contains at least one equivalent semiconductor device called anActiva. The most common circuit form used in the neural system employs the Activa in a groundedbase (podite) circuit. A synapse is also a circuit containing an Activa connected to form an electricaldiode. These configurations and their performance are discussed in detail inwww.hearingresearch.net/pdf/4Physiology1.pdf Only the outer and inner hair cells use a morecomplicated circuit. The circuit of these cells is discussed inwww.hearingresearch.net/pdf/5Generation.pdf

1.4.1.2 The bandwidth and propagation velocity of myelinated neurons

The literature contains a variety of comments concerning the propagation velocity of neural signals. Little justification is given for these comments except for a general expression that velocity is afunction of the diameter of the axon.

It is very important to distinguish between the actual diameter of the axon of a myelinated neuronand the overall diameter of the axon/myelination combination. The actual diameter of most neuronsvaries very little. However, the thickness of the myelination can be so great as to vary the diameterof the combination by a factor of ten. It is the reduction in the capacitance per unit length of the axondue to the thickness of the myelination that determines the propagation velocity of the axon. Thissubject is addressed in detail at www.4colorvision.com/neuron/pdf/6Electrophysiol.pdf

The morphology of a signal propagation neuron, includes a series of Nodes of Ranvier along thelength of its axon. Each Node of Ranvier acts as a regenerative repeater of the signal traveling alongthe axon. The repeater introduces a significant time delay. As a result, the average propagationvelocity of a signal projection neuron is dominated by the sum of the repeater delays. As a result, thenominal (first order) neural projection velocity over lengths longer than two millimeters is 44meters/sec. This number is much smaller than the 4400 meters/sec achieved by the intrinsic axonsegment between two regenerators.

The bandwidth of the axon is largely determined by the maximum clock speed of the action potentialsgenerated by the projection neuron and its Nodes of Ranvier. For most projection neurons, thisbandwidth is about 250-300 Hz. For some projection neurons between the spiral ganglia and theinferior colliculus, this bandwidth approaches 600 Hz in many mammals. There are indications thatsimilar circuits in the dolphin may achieve 900-1000 Hz. This would be an adaptation to optimizethe primary echolocation servo (PES) loop and provide the best possible biosonar range resolution inthe impulse part of the system.

1.4.2 The acoustic receiving & signal generation system in dolphins

1.4.2.1 Technical background & definitions

Dolphin hearing researchers have developed a term of art that is poorly defined in the literature. They compare dolphin hearing with an “energy detector.” In electrical engineering, a detector comes

5 September 200975

in a variety of forms. Some forms are envelope sensitive, some forms are phase sensitive and someforms are frequency sensitive. In photosensitive systems, the detector is either thermal energysensitive or photon flux sensitive. Thermal energy detectors are usually thought of as sensitive to abroad frequency range of radiant energy without regard to the phase of the individual frequencycomponents of that energy. Energy detectors typically have very long time constants and do notrespond to changes rapidly. This includes changes in the envelope of the applied signal. From thecontext, it appears dolphin researchers imply this form of energy detector. However, there are clearexamples where this is not the type of detector used by the dolphin. The phase sensitivity of thedolphin signal detectors will be discussed below.

Dispersion is a term of art that has not appeared previously in dolphin research related to thecochlea. Acoustic dispersion is the spreading of energy associated with an initial signal in either thetemporal or frequency domain. Temporal dispersion of a longitudinal acoustic wave is foundtheoretically in both Newtonian and non-Newtonian fluids. The result is different velocities forfrequency components as a function of their frequency. Both fresh and salt water at temperaturessignificantly above their freezing point are Newtonian fluids but they show negligible dispersion inthe context of hearing. This is not true of silt laden water. Silt laden water is highly non-Newtonian,exhibits measurable temporal dispersion, but is of little interest here.

Frequency dispersion of acoustic energy may occur in analogy with light passing through a prism. Inthe cochlea, the energy associated with an initial signal travels as a surface acoustic wave alongHensen’s stripe on the surface of the tectorial membrane. As the curvature of Hensen’s stripe varies,the energy is dispersed spatially across the surface of the tectorial membrane.

1.4.2.2 Details of acoustic energy reception in the dolphin

Acoustic antenna rules would suggest the outer ears of dolphins perform in two different modesdepending on the frequency of operation. At low frequency, the wavelengths of signals in water arelonger than the individual elements of the receiving system. This suggests that at low frequencies,dolphin hearing is largely omnidirectional. It also suggests, the outer ear is merely an impedancematching device between the water and the entrance to the cochlea of the inner ear. At frequenciesabove 15-30 kHz, the wavelengths of sound are much shorter. In this higher range, the receiving(and transmitting) elements of the dolphin support highly directional antennas.

The literature does not provide a detailed description of the outer ear of the dolphin when operatingat high frequency. It only notes the outer ear is based on a graded index form of dielectric antenna,instead of the horn type of antenna used by other mammals. The dielectric form allows ahydrodynamic exterior surface that is compatible with a forward looking receiving system yet meetsthe requirements of high speed swimming. The problem in defining the dolphin antenna is twofold. Insufficient detail is available concerning the fatty tissue material along the lower posterior jaw ofthe dolphin and it is quite likely, the precise geometric form of the fatty material is under themuscular control of the dolphin. While the figure in Section 1.1.3 shows the dolphin outer ears asviewed from the front, many other pictures and drawings suggest significantly different externalappearances for the outer ears (See Reynolds, 2000).

1.4.2.2.1 A physiologically compatible lens antenna for reception

A simple description of the external ear of the bottlenose dolphin is shown in Figure 1.4.2-1based onthe lower half of the head of the dolphin in [Figure 1.2.2-2 ] from Norris. The fatty tissue deposits

5 September 200976

appear to form a lens-type antenna between the vestibule of the labyrinth and the exterior waterenvironment. The fatty deposits are known to vary in density. Hence, the lens-type antenna is of thevariable density gradient type. A variable density contributes to a shorter antenna for a given degreeof quality (minimizing sidelobes in the far-field pattern for a given aperture size).

The goal of any lens-type receiving antenna is to deliver energy to the receiving circuitry at a higherpower density than present at the aperture of the antenna. A lens-type antenna is not as frequencysensitive, within the limits of geometrical optics, as most types of antennas . The design offersminimum sidelobe sensitivity, as observed in practice. The angular size of the main lobe is based onthe aperture of the antenna and the wavelength of the energy involved.

The delivery of the acoustic energy to a thin section of the vestibule of the labyrinth explains one ofthe unknowns in earlier dolphin research, how the auditory system worked with the obviouslyatrophied and disconnected bones of the middle ear. As noted earlier, an aquatic animal has no needfor the impedance conversion function provided by these bones in the terrestrial middle ear.

The “squint angle”between the two ears , a term adopted from radar design, is shown as 20 degreesin the horizontal plane. This is the value used by Au & Moore as nominal. A feature of the variable

Figure 1.4.2-1 The outer ear and field patterns of the bottlenose dolphin. The outer ear is a variable density acousticlens formed of the fatty deposits at the rear of, and alongside, the lower jawbone. The outer ear acts like a Galileantelescope, transforming a low energy density planar wavefront at the aperture into a higher density wavefront ofgreater curvature at the entrance to the labyrinth and inner ear. The energy delivered to the labyrinth is transferred tothe cochlear partition where it is processed conventionally. The beam pattern is nominally 20 degrees between nullsand is independent of acoustic frequency. The horizontal “squint angle” between left and right ears is shown at 20degrees.

5 September 200977

97Goodson, A. & Klinowska, M. (1989) A proposed echolocation receptor for the bottlenose dolphin(Tursiops truncatus) In Thomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans, Vol 196. NY:Plenum pp 255-267

98Dobbins, P. (2007) Dolphin sonar–modelling a new receiver concept Bioinspiration & Biomimetics vol 2,pp 19-29

99Wever, E. McCormick, J. Palin, J. & Ridgway, S. (1971) The cochlea of the dolphin, Tursiops truncatus:general morphology Proc Nat Acad Sci USA vol 68(10), pp 2381-2385

100Wever, E. McCormick, J. Palin, J. & Ridgway, S. (1971) The cochlea of the dolphin, Tursiops truncatus:the basilar membrane Proc Nat Acad Sci USA vol 68(11), pp 2708-2711

density lens-type external ear is the ability to change the squint angle easily under muscular control. Under optimum hydrodynamic drag conditions, the squint angle can be larger and provide a broadersearch capability of low resolution. The squint angle can be reduced during an attack phase toachieve greater angular precision in target tracking and identification but at some loss inhydrodynamic efficiency.

1.4.2.2.2 The potential for the teeth to form an antenna in dolphins

Goodson & Klinowska proposed in 1989 that the equally spaced teeth of the bottlenose dolphin T.truncatus might form an antenna array providing a narrow forward-pointing beam supporting thehearing modality97. The proposal was primarily conceptual. It was based on certain anatomicalobservations concerning the teeth and offered no details of the relevant physiology of the dolphin. No comprehensive discussion of potential antenna designs was presented. Dobbins recentlypresented a design review of the Goodson & Klinowska antenna proposal based on the academicstudies of two of his students98. While providing a competent analysis of the performance of a beamforming array of elements, it did not discuss the relevance of such an array to the physiology orneurology of the dolphin (page 27). Neither did it address one of the most frequent criticisms of theapproach, the ability of dolphins who have lost their teeth to achieve apparently normal echolocationperformance. A second argument against the approach is that the hardness of the teeth isincompatible with the absorption of acoustic energy from the environment or the soft tissue of thejaw.

1.4.2.3 Details of the frequency selection mechanism in the dolphin

The energy acquired by the outer ear is delivered to the inner ear and the cochlea as in all mammals. The cochlea of the dolphin is remarkably similar to that of humans in size and internal organization. It operates under the same principles as the human cochlea. The variable curvature of the criticalelement of the mechanism, Hensen’s stripe, causes the acoustic energy to be dispersed as a functionof frequency without the use of any resonant circuit elements.

Wever, et. al. presented a series of four papers on the anatomy of the cochlea in 1971-72. The last inMarch 1972 (referenced below) provides references to the earlier papers. They showed the cross-section of the cochlea is very similar to that of other mammals, except the individual elements aremuch more rigid99. Figure 1.4.2-2 described the curvature of the cochlea in this species compared tothat of human100. Using the theory of this work, the dolphin shows much higher discriminationcapability in the high frequency region because of the low change in curvature with rotation in theouter portion of the cochlear partition. While one might assume the shorter cochlear partition wouldlead to a more limited acoustic range than humans, this is misleading. It is the maximum and

5 September 200978

101Wever, E. McCormick, J. Palin, J. & Ridgway, S. (1972) Cochlear structure in the dolphin,Lagenorhynchus obliquidens Proc Nat Acad Sci USA vol 69(3), pp 657-661

102Ketten, D. & Wartzok, D. (1989) Three-dimensional reconstructions of the dolphin ear In Thomas, J. &Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pp 81-105

103Ketten, D. (1992) The cetacean ear: form, frequency, and evolution In Thomas, J. Kastelein, R. & Supin,A. eds. Marine Mammal Sensory Systems. NY: Plenum pp 53+

minimum rate of change in curvature that determines overall frequency range of a species. Thechange in curvature with rotation in the inner ear is less than in humans over much of its range. The human cochlear partition is known to exhibit an increased curvature near it’s apical (smalldiameter) terminus that has been labeled a “hook.” The dolphin cochlear partition also appears toexhibit such a hook.

Wever et al. (1972, p.659) discuss the variationsin the uniformity of curvature in a relateddolphin species, the Pacific white-sided dolphin,Lagenorhynchus obliquidens101. They notesignificant variations in the curvature amongspecimens which they relate to growthanomalies. While probably true, theseanomalies would result in significantdifferences in local frequency discriminationcapability in these specimens. Thesedifferences would probably not be recognizablein any but the most sophisticated evaluationprograms. A variety of parameters in L.obliquidens and T. truncatus are compared bythese authors.

Ketten & Wartzok have also studied the cochleaand provided dimensions of the cochlearpartition102. They describe the shape of the cochlear partition of Delphinidae as a gnomonic(logarithmic) or equiangular spiral based on the “lateral edge” of the cochlear partition. While theirdata captures a variety of ratios related to the physical parameters of the cochlear partition, they donot adequately capture the geometry of the critically important tectorial membrane. It appears theirlateral edge is the outer edge of the partition rather than the inner edge adjacent to Hensen’s stripe. The use of the inner edge would introduce some differences in their data. They define a type I andtype II spirals. The type II, equiangular (or logarithmic) spiral is only a first order approximation tothe modified Hankel function actually used in the dolphins and other mammals. The modifiedHankel function is discussed in a following section. Their type I Archimedean spiral appears to be aperturbation on the type I of this work.

Ketten provided a graphic of the cochlear spiral of Delphinidia in 1992103. Figure 1.4.2-3 shows herfigure. The change in the rate of change of curvature of the cochlear partition near the indicatedmark allows the family to achieve higher frequency resolution in the higher frequency range withoutsignificant loss in resolution at lower frequencies.

Figure 1.4.2-2 A comparison of dolphin and humancochlear curvatures drawn to the same scale. The dolphin(a) shows much lower curvature per unit rotation in thehigh frequency outer region but a shorter cochlearpartition and higher curvature per unit rotation in the lowfrequency region. From Wever et al., 1971.

5 September 200979

104Lemonds, D. Au, W. Nachtigall, P. Roitblat, H. & Vlachos, S. (2000) High frequency auditory filtershapes in an Atlantic bottlenose dolphin J Acoust Soc Am vol 108(5), Pt 2, pg 2614

105Wever, E. McCormick, J. Palin, J. & Ridgway, S. (1971) The cochlea of the dolphin, Tursiops truncatus:hair cells and ganglion cells Proc Nat Acad Sci USA vol 68(12), pp 2908-2912

Two other important factors related tofrequency discrimination have been reported. As found in other mammals, the dolphinexhibits a critical bandwidth. This criticalbandwidth can be considered the effective noisebandwidth of the animals receiving system. Alternately, it can be considered the bandwidthover which the animal is able to correlatefrequency information as part of theinformation extraction process within the brain. The critical bandwidth of the dolphin appears tobe similar to other mammals. Lemonds, et. al.have provided values for the critical band atseveral frequencies104. They report 16% ofcenter frequency at 40 kHz to 11% of centerfrequency at 100 kHz.

The critical bandwidth figure, and the large sizeof the medial geniculate nuclei in the dolphinmake it very likely that the dolphin employs thesame spatial correlation mechanisms asemployed by humans in both the visual andauditory domain. This would suggest thedolphin can in fact “image” targets examined byits active sonar system at a quality level yet tobe determined. The limiting resolution of thisimaging capability would be on the order of one-quarter wavelength or about 0.25 cm at 150 kHz. Ifcorrect, future exploration of the diencephalon of the dolphin brain should focus on describing thereticulated portion(s) of the very large (relative to human) medial geniculate nuclei. It may bepossible to locate the portions described as the perigeniculate nuclei in this work (Section 1.3.3).

Wever, et. al. of 1971 (p. 2911) compare the number of hair cells and ganglion cells of the dolphin andhuman cochlea105. While the number of hair cells is remarkably similar to humans, the number ofganglion cells differs considerably. The bottlenose has about three times as many ganglion cells asthe human.

Some of the axons in the auditory nerve of the dolphin are known to be quite large. Size in this areaallows the transmission of wider bandwidth action potential pulse streams. A few terrestrial animalsare known to project signals over the stage 3 signal projection circuits leading to the inferiorcolliculus at up to 600 Hz. It is quite possible that the dolphin can project signals over these circuits

Figure 1.4.2-3 The “cranked” form of the cochlearpartition in Delphinidia. The change in the rate of changeof curvature near the mark is difficult to see. The changeaccounts for the heightened frequency resolution ofDelphinidia at high frequencies. From Ketten, 1992.

5 September 200980

106Zook, J. & DiCaprio, R. (1989) A potential system of delay-lines in the dolphin auditory brainstem InThomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pg 190

107Ketten, D. (2000) Cetacean ears In Au, W. Popper, A. & Fay, R. eds. Hearing by Whales and Dolphins.NY: Springer Chapter 2

up to at least 1,000 Hz and possibly 3,000 Hz106. If this is confirmed by more detailed experiments, itwould mean the dolphin could employ its two primary auditory channels as phase coherent receiversup to 1,000 Hz. Such operation would improve its horizontal spatial angular resolution considerablyover that achievable by other mammals.

1.4.2.4 Details of neural signal generation in the dolphin

The generation of neural signals in the dolphin follows the general plan of mammals. The shape andlength of the cochlea and the arrangement of the hair cells are remarkably similar to that of humans. Ketten has tabulated the parameters of the cochlear partition for a variety of animals107. Theprincipal deviation between the bottlenose dolphin and the human is in the specific curvature ofHensen’s stripe in the cochlea. This deviation accounts for the difference in frequency range anddifferential frequency sensitivity of the dolphin cochlea compared to other species.

The morphology of the cochlea of the dolphin is virtually indistinguishable from that of the human. Therefore, available caricatures from the human in Chapter 4 of “Processes in Biological Hearing”can be used here effectively. Figure 1.4.2-4 shows the typical method of signal generation inmammals. Cartoons describing this mechanism are available athttp://hearingresearch.net/anim/extraction.htm and http://hearingresearch.net/anim/separation.htm .

5 September 200981

Figure 1.4.2-4 The acoustic information extraction mechanism of hearing. A; the application of acoustic energy tothe SAW filter of the tectorial membrane. The perilymph filling the scala vestibuli and the scala tympani does notparticipate in the primary acoustic processes. The primary mechanisms are immersed in the endolymph. Thelauncher portion of Hensen’s stripe is excited within the vestibule of the labyrinth by energy propagated from thestapes and oval window. B; the transfer of the extracted energy to the neural system. Two separate types of signalsare generated by the inner and outer hair cells. When coiled, the energy transfer into these channels is a function ofthe curvature of Hensen’s stripe.

5 September 200982

1.4.2.4.1 The dispersion of the received energy within the cochlea of dolphins

As developed in Chapter 4 of “Processes in Biological Hearing,” the nominal curvature of Hensen’sstripe is given by a mathematical formula known as a Hankel function. This function describes thecomplete frequency-position-delay characteristic for a given mammalian species. As noted earlier,many mammals exhibit an increase in curvature (decrease in local radius) near the apex of thecochlea. The dolphin appears to be one of these species. Figure 1.4.2-5 illustrates a typical cochlearpartition from a bottlenose dolphin overlaid by a Hankel function. The parameters of the Hankelfunction are Start = 2.0, End = 15.4 and Hook parameter = 18. The figure introduces something of anoptical illusion due to the variation in the width of the solid line representing the cochlear partition. The dashed line is a simple mathematical function with a smoothly varying curvature. It is thecochlear partition that is somewhat irregular.

The agreement between the measured shape ofthe cochlear partition from a single bottlenosedolphin without range bars and the theoreticalHankel function appears excellent. Thevariation is well within the range found in otherspecies, including humans. The variationsuggests the animal had some deviations fromwhat one might call “hearing with a linearvariation in frequency with position” If oneplots the local radius of curvature (anapproximation to the true curvature of aHankel function) and scales it properly, oneobtains the frequency place characteristic forthis species. Figure 1.4.2-6 shows thischaracteristic, using the conventional scales ofhearing research log-log scales, for the nominalbottlenose dolphin cochlea described by theabove Hankel function. The abscissacorresponds to the distance along the cochleafrom its initial point within the vestibule of thelabyrinth to its terminus at the helicotrema. The ordinate scale represents the frequencyrange to which the species is sensitive. Theprecise center frequency of the first and last setof outer hair cells in the cochlea of the dolphin are not known. However, a scale from 150 Hz to 150kHz is frequently assumed for the bottlenose dolphin. These limits do not necessarily correspond tothe limits of the sensitivity spectrum shown in the next section. This figure also shows the relativedelay to be expected, assuming the velocity of propagation along the stripe is six meters/sec. Thisdelay is important in describing the neural network of the dolphin later in this work.

Figure 1.4.2-5 A typical dolphin cochlear partitionoverlaid by a Hankel function. See text. Dashed curve isHankel function for 2<x<15.4. Solid curve fromMcCormick, 1971.

5 September 200983

Figure 1.4.2-6 Performance parameters for Hensen’s stripe for the nominal dolphin cochlea.

5 September 200984

108Supin, A. & Popov, V. (1989) Frequency-selectivity of the auditory system in the bottlenose dolphin,Tursiop turncatus in Thomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pp385-393

109Johnson, S. (1967) Sound detection thresholds in marine mammals In tavolga, W. ed. MarineBioAcoustics NY: Pergamon Press pp 247-260

1.4.2.4.2 Initial signal generation by phonoreceptors in the dolphin

The generation of the initial signals by the hair cells of the dolphin cochlea can be assumed to be likethat of all other mammals explored to date.

The arrangement of both the inner and outer hair cells along the basilar membrane and their contactwith the tectorial membrane is virtually identical to that found in the human. As a result, both thetemporal signals associated with the inner hair cells and the tonal signals associated with the outerhair cells can be expected to be generated in the same manner.

1.4.2.4.3 Energy vs phase sensitivity of the signal generation mechanism

Some data is available that limits consideration of the dolphin receiving system as an energydetector. In addition, the binaural sensitivity of the dolphin at low frequency, like that of the human,is not compatible with a simple energy detector.

Supin & Popov have reported on the sensitivity of the receiving system to the variation in the shape(based on phase changes) of the envelope of individual pulses but on a global basis108. The globalbasis involved the measurement of evoked potentials measured at the surface of the animals head. They assert these signals can be described as auditory cortical responses due to their similarity toauditory brainstem responses. Strictly speaking, such phase sensitivity is not compatible with anenergy detector.

The performance of the tonal and temporal channels of hearing should be examined separately, andthen from a binaural perspective, to define the phase performance of the overall system adequately.

Chapter 5 of “Processes in Biological Hearing” describes the phase sensitivity of the sensory neuralchannels at frequencies below 600-3,000 Hz depending on the species. Below this frequency, thebinaural signals projected to the inferior colliculus are compared on a phase basis. Below thisfrequency, the detection system can only be considered an energy detector on a very short term basis. Above this break frequency, the sensory neurons operate as nearly perfect integrators, and can thusbe described as energy detectors.

1.4.2.5 Composite performance of the dolphin

Several performance parameters of the dolphin can be deduced from only the above discussion of onlythe physiological and neurological elements of its hearing.

1.4.2.5.1 The limiting auditory sensitivity of dolphins

Figure 1.4.2-7 compares the acoustic sensitivity of the dolphin and the human109. While the absolutesensitivity is higher due to the larger collection apertures of its “outer ears,” it also achieves a

5 September 200985

considerably higher ultimate frequency.

1.4.2.5.2 The frequency discrimination capability of the dolphin

Based on the geometry of the cochlea in the dolphin, the frequency discrimination capability of thecochlea is fmax/fmin = (min. step)3500 before considering any rolloff in frequency response due to otherfactors. For fmax = 150 kHz and fmin = 200 Hz, the minimum step is 1.0019 or slightly less than twotenths of 1percent (~0.2%) of the characteristic frequency. This value is 28.5 Hz at 15,000. Thisvalue will be reduced by any diffraction within the dispersion mechanism and the finite size of theouter hair cells. It may also be reduced by any longitudinal summation of the signals from adjacentouter hair cells. It could be increased by interpolation within the signal manipulation process butthis is unlikely because of the automatic gain control built into the sensory neurons. This percentagevalue is very close to the mid band values reported by Au (1993, p. 41) and shown in Figure 1.4.2-8.

Figure 1.4.2-7 Sensitivity of dolphin compared to human hearing. The difference in absolute sensitivity is related tothe larger receiving aperture of the outer ear. From Johnson, 1967.

5 September 200986

The above discussion strongly suggests the whistles of the bottlenose dolphin are due to their use of aswept continuous tone Doppler radar for echolocation in the range of 100 to 600 meters and that thissystem achieves near theoretical performance within this range.

A problem appears in the 1993 book by Au. The apparent excellent frequency discriminationprovided by the cochlea stands in stark contrast to Au’s statement on page 122 based on a

Figure 1.4.2-8 Frequency difference threshold for the bottlenose dolphin and humans. Theoretical limit is based onthis study and applies to the swept continuous tone echolocation capability (generally below 30 kHz). Other datafrom Thompson & Herman, 1975.

5 September 200987

110Thompson, R. & Herman, L. (1975) Underwater frequency discrimination in the bottlenose dolphin andthe human J Acoust Soc Am vol 57, pp 943-948

paper by Thompson & Herman110. Au claims the bottlenose dolphin cannot perceive frequencydifferences of better than 8 kHz in the 100 to 130 kHz range. The problem appears tooriginate in the reading of figure 2 of Thompson & Herman. The ordinate scale includes afactor of 103. However, the scale is a ratio relative to 100,000 Hz. The absolute value of thedifference limen expressed as a differential frequency threshold is 800 Hz (in agreement withTable I of the Thompson & Herman paper). Au actually reproduces figure 2 of Thompson &Herman on page 41 and draws the correct conclusion that the limiting limen is near 0.2% (285Hz at 15 kHz), rising to near 800 Hz near the edge of the effective operating range.

1.4.3 Signal processing & signal propagation in dolphins

This section stands as a placeholder. Virtually nothing is known about the signal processingperformed in the neural system of the dolphin between the signal generating phonoreceptors of thecochlea and the CNS, other than some counts of the number of neurons exiting the spiral ganglia. Signal propagation in the dolphin appears analogous in every way with that discussed in Chapter 7of “Processes in Biological Hearing.”

The spiral ganglia of the dolphin is homologous with that of other mammals. No references werefound in the literature detailing the spiral ganglia adjacent to and intertwined with the cochlea. It isknown to create about three times the number of ganglion output cells as found in the human spiralganglia. Precisely where these neurons extend to within the cochlear nucleus is unknown in 2006. The general physiology associated with the spiral ganglia can be found in Chapter 6 of “Processes inBiological Hearing.”

Very little is known about the neural system of the dolphin at the detailed level. The most prominentfeature is the physical cross section of some of the neurons emanating from the spiral ganglia. Whilethe literature has asserted that this would make the signal propagation somewhat faster based onthe chemical theory of the neuron, the situation is more complex. Based on the electrolytic theory ofthe neuron, underpinning this work [Chapter 14 of “Processes in Biological Vision,”www.sightresearch.net/pdf/14Tertiary.pdf ], both the bandwidth and the average propagation velocityof neurons are determined primarily by the characteristics of the Nodes of Ranvier that occur alongthe length of propagation neurons. The diameter of an axon is a secondary factor. The highestbandwidth neurons previously reported in mammals are found in the neurons involved in the phasecomparison circuits used in defining the direction to a source and possibly in neurons of the precisionoptical servo system (POS). This maximum is near 600 Hz. It is possible some of the neural pathsemanating from the spiral ganglia neurons in the dolphin could exhibit bandwidths up to about 1,000Hz may be present. The actual value of this parameter is important in delimiting the performance ofthe echolocation circuits of the dolphin discussed below.

The very general comments predicting the size of the axons in various neural paths based on timedelays encountered within the system do not reflect a detailed understanding of the neural system. Much of the delay encountered at different locations within the brain is due to the time allotted tosignal manipulation in one or more of the neural engines (generally described as independent nucleior areas of the cortex surface in the morphology literature) rather than due to variations in size of theinterconnecting signal projection neuron.

5 September 200988

111Ridgway, S. (2000) The auditory central nervous system of dolphins In Au, W. Popper, A. & Fay, R. edsHearing by Whales and Dolphins. NY: Springer pg 279

112Fulton, J. (2004) Biological Vision: A 21st Century Tutorial. Victoria, BC, Canada: Trafford. ISBN141201917-6

113Fulton, J. (2006c) Hearing: A 21st Century Paradigm. Victoria, BC, Canada: Trafford, ISBN 142516065-4

114Zook, J. Jacobs, M. Glezer, I. & Morgane, P. (1986) Some comparative aspects of auditory brainstemcytoarchitecture in echolocating mammals In Nachtigall, P. & Moore, P. eds. Animal Sonar: Processes andPerformances. NY: Plenum Press pp 311-316

Ridgway has discussed briefly the diameter of the neurons emanating from the spiral ganglia111. Hereports those between the spiral ganglia and the lateral lemniscus to be as large as 12 microns in thebottlenose whale, Hyperoodon ampullatus, with diameters of at least seven microns in the bottlenosedolphin.

1.4.4 Signal manipulation and cognition in dolphins

Some excellent studies of the anatomy of the dolphin brain have been published. However, very littlework has been published on the morphology of the brain in dolphins compared to that for bats.

Only brief physiological studies have been made into the neural system of the dolphin. Most of thedata has been obtained with non-invasive procedures, such as auditory brainstem responses (ABR). The analysis of the resulting data has been quite limited, mostly to comparing waveforms with thoseof other species.

The analysis of waveforms such as those by Ridgway in Au, Popper & Fay (p. 286) and inLeatherwood & Reeves (p. 91) can proceed much farther if the origins of the various components ofthe auditory brainstem response (ABR) are better understood. The reader is referred to Fulton(2005, Sec 11.1.6) for additional information on this subject. The information is applicable to both thevisual and auditory sensory domains.

This section will assume the neural system of the dolphin is typical of other mammals with theprimary differences being in ecologically defined optimizations in specific portions of the brain. As anoften quoted example, the eye sight of dolphins cannot be relied upon in murky water. As a result,the dolphin has optimized its phonic and auditory capabilities to provide an alternate high spatialresolution capability. The visual portions of its brain, both occipital lobe and the visual part of thediencephalon appear less developed than in humans. It has been reported that the dolphin has nofovea in its retina. However, this statement needs further confirmation.

Nearly all of the studies related to the neural system and brain of the dolphin must be considered asstatistically inadequate and only representing a unitary or very small sample group.

The model presented in Biological Vision112 and the similar model presented in Biological Hearing113,both by Fulton, suggests a framework for analyzing both the passive and active auditory capability ofthe dolphin. The figure in Section 1.1.2 above will be expanded in the following discussion.

1.4.4.1 The gross morphology, homeostasis and vascular circulation

Zook, et. al. have reviewed the available morphological material up to 1986114. Their discussion does

5 September 200989

115Morgane, P. Jacobs, M. & Galaburda, A. (1986) Evolutionary apects of cortical organization in thedolphin brain In Bryden, M. & Harrison, R. eds. Research on Dolphins. Oxford: Clarendon Press pp 71-98

116Morgane, P. & Glezer, I. (1990) Sensory neocortex in dolphin brain In Thomas, J. & Kastelein, R. eds.Sensory Abilities of Cetaceans. NY: Plenum Press pp 107-136

117Morris, R. (1986) The acoustic faculty of dolphins In Bryden, M. & Harrison, R. eds. Research onDolphins. Oxford: Clarendon Press chap 18

118Houser, D. Finneeran, J. Carder, D. et al. (2004) Structural and functional imaging of bottlenose dolphin(Tursiops truncatus) cranial anatomy J Exp Biol vol 207, pp 3657-3665

119Ridgway, S. (1990) The Central Nervous System In Leatherwood, S. & Reeves, R. eds. The BottlenoseDolphin. NY: Academic Press pp 70-77

120Ridgway, S. (2000) The auditory central nervous system of dolphins In Au, W. Popper, A. & Fay, R. edsHearing by Whales and Dolphins. NY: Springer pp273-293

121Ridgway, S. 1995) Dolphin Doctor, 2nd Ed. San Diego, CA: Dolphin Science Press pg 47

not progress beyond the concept of a filter as a black box. Some cytology of the brain has beenpublished but it is mostly in superficial form and not tied closely to the physiology of the system. Morgane, Jacobs & Galaburda, in chapter 5 of Bryden & Harrison, have provided some informationon the cytoarchitecture of the brain of Cetacea115. Their focus was on the neocortex to the exclusion ofthe diencephalon and other elements of the paleocortcx. They note the difficulty in comparing thebrains of Cetacea with Hominoidea because of the distinctly different ecological niches and resultantneural evolution. They suggest reverting back down the phylogenic tree to find a common ancestorand performing comparisons beginning from that base.

They do note, “since from the outset it is clear that cortical architecture in the primate brains isstrikingly different from that of the whales” (p. 84). The differences appear to relate to both thelaminar organization of the neocortex and the cell populations within those layers. The neocortex isprobably the least important part of the bottlenose dolphin brain (except for the frontal lobe) whendiscussing echolocation. Morgane & Glezer have also written on the morphology of the neocortex116. They went farther and discussed the cytology of some areas of the neocortex.

Morris has noted the unusual vascular system in dolphins and its potential utility in connection withdeep diving as well as thermal control and possible dampening of vascular pressure variations117. Houser, et. al. have performed a broader set of in-vivo studies using medical imaging equipment toprovide significantly more information in this area118. The studies provided significant newinformation in several areas. The role of the vascular circulation in maintaining uniformtemperature within the fat tissue representing the beam forming elements of both the transmittingand receiving portions of the echolocation system was emphasized. The role of the air sacs insurrounding the bulla of the ears and providing acoustic isolation from the nearby transmittingelements was emphasized. Additional details concerning the pneumatic pathways within the headwere also provided.

Ridgway has also provided brief discussions of the brain of the bottlenose dolphin119,120. He also notesthe apparent ability of the animal to achieve a sleep-state in only one-half of the brain at a time. Although able to sleep with both eyes closed121, the animal is known to enter a sleep-state with oneeye remaining open at a time. Ridgway did not discuss the importance of the paleocortex toecholocation and other specialized capabilities of the dolphin. Ridgway did note the fact the pinna is

5 September 200990

122Bullock, T. & Gurevich, V. (1979) Soviet literature on the nervous system and physiology of Cetacea Internat Review of Neurobiol vol 21, pp 49-109

lacking in dolphins and the earlier comment of McCormick that the malleas of the middle ear doesnot attach directly to the tympanic membrane. Unfortunately, the above presentations did not describe the paleocortex and neocortex or presentimagery showing the posterior dorsal aspect. Bullock & Gurevich have provided a broader range ofdetails of the CNS of the bottlenose dolphin based on early studies in the Soviet Union122. Theyhighlight the significant differences in size between the elements of the paleocortex in the dolphin,the bat, and in humans. They note particularly the “prodigiously large” size of the lateral lemniscusof dolphins, including a nucleus unknown in humans. They make similar observations concerningthe inferior colliculus (p. 62). Beginning on page 59, they discuss aspects of both the dorsal andventral cochlear nuclei.

Figure 1.4.4-1 reproduces a crucially important table derived from Zvorykin by Bullock & Gurevich. It has been expanded to stress the massively larger acoustic elements of the brainstem of the dolphincompared to the human, on both an absolute volume basis and relative volume basis using theweights in column 15. This contrasts with the visual components shown that are only marginallylarger. It also contrasts with the relatively smaller cerebral cortex of the dolphin compared to thehuman (not shown).

The proposed activity associated with each of these components is shown below the tabulation. Except for echolocation, these labels are derived from studies of the human and terrestrial mammals.

Figure 1.4.4-1 Volumes of auditory and visual structures in the brain of dolphins (Delphinus delphis), bat (Pipistrellus pipistrellus), and man (Homo sapiens) in cubic millimeters. Based on Bullock & Gurevich, 1979.

5 September 200991

123Zook, J. & DiCaprio, R. (1989) A potential system of delay-lines in the dolphin auditory brainstem InThomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press

The proposal that the primary active echolocation signal manipulation area is the lateral lemniscusis based on its massive proportions and its vestigial character in humans.

No detailed discussion of the morphology of the lateral lemniscus was found in the literaturealthough Zvorykin did separate it into three engines in his tabulation. It is suggested that thismassive neural engine consists of many more engines, specifically a thin reticulated outer layerhomologous with the dolphins PGN (and possibly the more inclusive TRN) and a massive volumetricinterior homologous with the dolphins pulvinar. The former can be described as a perilemniscus or apons reticulated nucleus (PRN), with the other engines of the lemniscus retaining their currentdesignations. This would result in an area of the pons with a strong morphological similarity to theMGN/PGN/pulvinar couple of the thalamus.

The comment is often made that the overall size of the dolphin brain is comparable to that of thehuman. It is not always noted that the total weight of the adult animal is on the order of three timesthe size of the human.

1.4.4.1.1 The internal architecture of the engines of the dolphin brain

Many authors have noted the differences in the laminations of the dolphin CNS and that of othermammals. Bullock & Gurevich referenced Bogoslovskaya (1974) concerning the important featuresof the dolphin dorsal cochlear nucleus. They referenced Avksent’yeva (1974) who noted the fourthlayer was not differentiated from the third and fifth layer in the sensory cortex.

1.4.4.2 The neural circuits of the dolphin paleocortex EDIT

As noted in Section 1.4.4.1, the paleocortex of dolphins is far more developed than that inHominoidea. However, it has not been extensively studied at the histological or cytological level.

1.4.4.2.1 The functional organization of the cochlear nucleus

Zook & DiCaprio have explored the possibility that the cochlear nucleus contains one or more sets ofdelay lines123. The role of delay lines can be illustrated more completely by examining the overallsignal processing system. As an example, the slow wave structure of the cochlear partitionintroduces a time delay between longitudinally sequential inner and outer hair cells. This delay isinherent in the place-frequency-delay characteristic of all cochleas. The resulting signals must eitherbe processed in this echelon time structure or be realigned to a common time base through the use ofdelay lines prior to additional processing. In the human and other mammals studied, the echelontime structure is used within the POS, and probably within the PES, to establish the direction totargets of interest. A delay, like that introduced by the cochlear partition, is actually introducedintentionally within the POS of the visual system. The delays are then removed for furtherprocessing.

Zook & DiCaprio describe one of the neural paths between the ventral cochlear nucleus (VCN) andthe medial nucleus of the trapezoidal body (MNTB). They note the difficulty of measuring thedistance precisely and estimate its path length as between 1.0 and 2.5 cm. They also measure whatthey consider the differential length of the neurons along this path. This differential path isassociated with the cells between the top and bottom of one set of neurons arriving at the MNTB.

5 September 200992

They report this distance varied between 3.4 to 4.5 mm in Stenella and 2.3 to 4 mm in Tursiops. While it is too early to define this set of neural paths functionally, they appear to be functionallyhomologous with Meyer’s Loop in the visual system of mammals.

In the visual system, all of the delays are generated within the system. However, in acoustic system,the speed of sound is relatively low and appreciable transit delays occur in the external environmentas well as in the neurological system, especially in the bidirectional sound travel inherent inecholocation.

Zook & DiCaprio have also discussed the massive size of the trapezoid body of the cochlear nucleus inthe dolphin. The cochlear nucleus is itself massive compared to that in the human. The dorsalportion is about 10 times larger in the dolphin. The ventral portion is between 15 and 24 timeslarger based on the table from Zvorykin (Section 1.4.4.1).

1.4.4.2.2 The functional organization of the lateral lemniscus (Reserved)

1.4.4.2.3 The functional organization of the inferior colliculus

The assumption is that the source location function is concentrated in the inferior colliculus of T.truncatus. The challenge is to describe the inferior colliculus required to achieve the experimentallymeasured performance of the dolphin. It is important to recall that the only signals received at theinferior colliculus associated with high frequency ranging are stage 3 pulses describing the time ofinitiation and the amplitude of the envelope of the energy received at the cochlea at variousfrequencies defined by the dispersion properties of the cochlea. The Hankel function used in thedolphin cochlea has been modified to achieve the required spectral segregation. While many havediscussed the absolute timing precision required to achieve the observed targeting performance indolphins, the important parameter is the relative timing precision between the two auditorychannels. As long as the two channels are well matched in performance, any absolute timing errorswill tend to cancel.

To achieve the performance described above, each inferior colliculus must be able to create a multi-dimensional correlation filter that can compare frames of data received over intervals on the order of25-50 ms, a number compatible with that found in the neurological system generally. Each frame ofdata must be organized in a three dimensional matrix describing the distribution of the signalsreceived as a function of azimuth, elevation and nominal center frequency. For the resolutionsdescribed above, the matrix must be on the order of 256 bins wide in azimuth, on the order of 32 binsin elevation and on the order of 60 bins in frequency resolution (100 Hz/bin). The 60 bins infrequency resolution is adequate to cover the entire velocity range of the dolphin during a feedingattack. Even small integer multiples of these numbers for the matrix size are quite achievable. During search operations, the amplitudes of the returns are not as important as the precise time ofsignal receipt for a particular frequency component. Thus the amplitude stored in the matrix shouldbe that derived from the ranging information, relative time of signal receipt relative to the average ofthe time of receipt of all signals within the frame time. During target selection, from within a group,the amplitude of the received signals is more important since the differential amplitude determinedbinaurally provides the azimuth information. Similarly, the differential amplitude determined bycomparing frames acquired during target illumination by alternate vocalization sources provides theelevation information.

The proposed dolphin inferior colliculi must each manipulate the incoming signals from the twoauditory channels in order to compare frames based on different vocalization sources and obtain

5 September 200993

elevation information. The inferior colliculi must manipulate the incoming signals to compare signalamplitudes during vocalization from either source to extract azimuth information. Each inferiorcolliculus must extract a mean time delay between the start of either vocalization and the receipt ofthe returning signals. It must also determine the relative time of receipt of each frequencycomponent compared to this mean time delay. By comparing the time of receipt of each frequencycomponent between frames acquired during the falling and the rising portions of an individualvocalization, it can calculate both the range to the individual reflection and the Doppler velocity ofthat reflection relative to the mean range and Doppler velocity of the total target ensemble. With theabove information, the dolphin is able to generate a perception of any target ensemble in multi-dimensions that include mean range and velocity of the ensemble as well as the differential velocity,differential azimuth and differential elevation of each individual component of the ensemble withinthe resolution capability of the system. The required matrix algebra is easily achieved using aninferior colliculus consisting of two portions, 1. a multi-dimensional correlator formed of multiple multilayer folds of neural tissue similar to thatof all elements of the mesencephalon derived from the thalamic reticular nucleus (TRN) and2. The necessary sensing and computational circuitry similar to that found in all TRN’s.

- - - -

5 September 200994

Figure 1.4.4-2 The characteristics of various reflected signals expected in biosonar applications. A; a continuousfixed frequency signal and its reflection from three targets. B; a continuous swept frequency signal and severalreflections. C; a pulse signal reflected from a target and clutter due to the background. See text.

5 September 200995

1.4.4.2.4 A conceptual inferior colliculus in the dolphin

The assumption is that the source location function is concentrated in the inferior colliculus of T.truncatus. The challenge is to describe the inferior colliculus required to achieve the experimentallymeasured performance of the dolphin. It is important to recall that the only signals received at theinferior colliculus associated with high frequency ranging are stage 3 pulses describing the time ofinitiation and the amplitude of the envelope of the energy received at the cochlea at variousfrequencies defined by the dispersion properties of the cochlea. The Hankel function used in thedolphin cochlea has been modified to achieve the required spectral segregation. While many havediscussed the absolute timing precision required to achieve the observed targeting performance indolphins, the important parameter is the relative timing precision between the two auditorychannels. As long as the two channels are well matched in performance, any absolute timing errorswill tend to cancel.

To achieve the performance described above, each inferior colliculus must be able to create a multi-dimensional correlation filter that can compare frames of data received over intervals on the order of25-50 ms, a number compatible with that found in the neurological system generally. Each frame ofdata must be organized in a three dimensional matrix describing the distribution of the signalsreceived as a function of azimuth, elevation and nominal center frequency. For the resolutionsdescribed above, the matrix must be on the order of 256 bins wide in azimuth, on the order of 32 binsin elevation and on the order of 60 bins in frequency resolution (100 Hz/bin). The 60 bins infrequency resolution is adequate to cover the entire velocity range of the dolphin during a feedingattack. Even small integer multiples of these numbers for the matrix size are quite achievable. During search operations, the amplitudes of the returns are not as important as the precise time ofsignal receipt for a particular frequency component. Thus the amplitude stored in the matrix shouldbe that derived from the ranging information, relative time of signal receipt relative to the average ofthe time of receipt of all signals within the frame time. During target selection, from within a group,the amplitude of the received signals is more important since the differential amplitude determinedbinaurally provides the azimuth information. Similarly, the differential amplitude determined bycomparing frames acquired during target illumination by alternate vocalization sources provides theelevation information.

The proposed dolphin inferior colliculi must each manipulate the incoming signals from the twoauditory channels in order to compare (bivocal) frames based on different vocalization sources andobtain elevation information. The inferior colliculi must manipulate the incoming signals to compare(binaural) signal amplitudes during vocalization from either source to extract azimuth information. Each inferior colliculus must extract a mean time delay between the start of either vocalization andthe receipt of the returning signals. It must also determine the relative time of receipt of eachfrequency component compared to this mean time delay. By comparing the time of receipt of eachfrequency component between frames acquired during the falling and the rising portions of anindividual vocalization, it can calculate both the range to the individual reflection and the Dopplervelocity of that reflection relative to the mean range and Doppler velocity of the total targetensemble. With the above information, the dolphin is able to generate a perception of any targetensemble in multi-dimensions that include mean range and velocity of the ensemble as well as thedifferential velocity, differential azimuth and differential elevation of each individual component ofthe ensemble within the resolution capability of the system. The required matrix algebra is easilyachieved using an inferior colliculus consisting of two portions,

5 September 200996

1. a multi-dimensional correlator formed of multiple multilayer folds of neural tissue similar to thatof all elements of the mesencephalon derived from the thalamic reticular nucleus (TRN) and2. The necessary sensing and computational circuitry similar to that found in all TRN’s.

By introducing sensing neurons, where each sensing neuron, interrogates one row of a plane of thematrix, the significant features of the overall matrix can be extracted for pattern recognitionpurposes by subsequent engines of the brain.

The above conceptual inferior colliculus has a broad range of capabilities that appear to exhibitfeatures commonly used in simpler species:

1. It contains the bilateral field merging feature needed in the lateral geniculate nucleus of vision tomerge the half-hemisphere images from the two eyes.

2. It contains the two-dimensional planar matrix needed in the perigeniculate nuclei of both visionand passive hearing to extract specific information for pattern recognition purposes.

3. It obviously contains the individual spectral channel comparison features needed by a simplerinferior colliculus to extract the appropriate interaural time difference data (on a global basis) andinteraural spectral difference data (on an individual frequency basis) required to locate the centroidof a stimulus ensemble (such as an orchestra) and the differential positions of sources within thatensemble (such as a specific instrument) based on tonal differences.

1.4.4.2.5 The functional organization of the perigeniculate nucleus

The topology of the medial geniculate nucleus (MGN) of the dolphin has not been elucidated to date. However, its size is indicative of its importance. It is likely that the portion of this body homologouswith the acoustic perigeniculate nucleus (PGN) in the human will make up a significant portion ofthe dolphin medial geniculate nucleus. The perigeniculate nucleus forms a multidimensionalassociative memory, not unlike those found in man-made associative computer processors. It is thePGN that begins the information extraction process leading to cognition of a scene in the human.

The signals delivered to the acoustic perigeniculate nuclei of dolphins appear to be functionallyequivalent to those delivered to the visual perigeniculate nuclei of the visual system in humans. Thiswould suggest the dolphin could “image” acoustic signals in a manner that is not significantlydifferent from what humans can do in the visual regime (except for the limitations imposed by thewavelength of the signaling mechanism and possibly background noise).

Figure 1.4.4-3 shows the concept of the auditory signal correlation and interp extraction mechanismproposed for the human auditory system expanded to support the additional capabilities of thedolphin. The literature asserts the dolphin has approximately the same number of outer hair cells asdo humans and the cochlea is very similar in length. However, the curvatures of the cochlea aredifferent. As a result, the frequency range to which the cochlea responds is different. This isillustrated in the figure by the different slopes of the place frequency-delay characteristics on the left. Also added to the figure are the nominal frequencies supported by the active wide beam echolocationcapability and the active narrow beam echolocation capability. A challenge is to determine how theseadditional capabilities are processed within the overall system.

5 September 200997

The signals from the longitudinally oriented cochlea of humans are transformed into a rectangularformat by simple computational anatomy during transfer to the perigeniculate nucleus. The signalsfrom the two cochleas are also probably merged into a single composite format prior to correlation(although this could occur subsequently with some loss in signal-to-noise ratio). The width of thearray in humans is one octave.

Conceptually, the signals deposited at the nodes of the correlator are sensed by sensing neurons withlong neurites contacting the columns of neurons as shown. The signals generated by these sensingneurons as a group form the first order interp extracted from the data. These interps are transferredsequentially to the pulvinar for further information extraction leading to a higher level interp. Theseinterps are matched to patterns previously stored in the pulvinar. A successful match results in thegeneration of a percept, a fully resolved perception of the information supplied from the auditorysystem.

At first glance, it would appear the processing is likely to be very similar for dolphin hearing. Percepts could be generated from both the passive auditory system and the active echolocationsystem from the same basic neural architecture. It is reasonable to expect some additional neuralinputs from the vocal tract describing the timing and possibly the frequency and temporalcharacteristics of the emitted sound. Alternately, the auditory system may be able to record a sampleof the outgoing acoustic signal and use that to compare to the return signal. This information wouldallow the reflectance characteristics of the target to be determined as a function of frequency andtime. Such frequency information would be similar to that associated with color in the visualcorrelation system. The timing differences would be very similar to that associated with

Figure 1.4.4-3 Potential organization of the auditory analysis function in dolphins expanded from the nominalhuman capability. The musical notation is left on the figure for illustration. The correlator width of dolphins maydiffer from the correlator width (and indirectly the musical scale) of humans.

5 September 200998

reverberation, and incidental signal reflections, in human hearing.

On the assumption that the auditory signals of the dolphin are transferred to the perigeniculatenucleus in the same manner, the question of the width of the array arises. It is likely that the arraywidth is also one octave in the dolphin as well. The octave organization has a particular feature inhumans. Frequencies related by the exponential harmonic series, f = (base frequency)@2n wheren=0,1,2 . . describe a single perceived tone that can be labeled by the notes of the musical scale. Theeven harmonics not represented by the exponential harmonic series and all odd harmonics generatedifferent individual frequencies that are not consonant with the above series.

All of the exponential harmonics fall within a single vertical column in the figure. Sensing usingvertically oriented sense lines is easily able to detect the note associated with the exponentialharmonics. The effect is even easier to comprehend using the Spiral Tonal Map, Figure 1.4.4-4.

5 September 200999

By changing the curvature of the cochlea, the dolphin could use this same mapping to generateacoustic “images” quite similar to those generated by the (visual) PGN in humans and other higher

Figure 1.4.4-4 The human Spiral Tonal Map. The figure shows how exponential harmonics (in this case startingfrom a base at 100 Hz at upper right) all generate the same perception of a tone. In this case, the tone is half waybetween G and G#. From Chapter 9 of Fulton, 2007.

5 September 2009100

124Ridgway, S. (2000) The auditory central nervous system of dolphins In Au, W. Popper, A. & Fay, R. eds. Hearing by Whales and Dolphins. NY: Springer Chapter 6

mammals with a foveola. An example of this capability is shown in Figure 1.4.4-5. Although we donot know the aspect ratio of the correlator in dolphin, the overall capability for the equivalent(hearing) PGN is not likely to be smaller. Thus, this figure gives a suggestion of the acoustic imagingcapability of the dolphin where the minimum pixel size is probably 15 to 45 arc minutes asdocumented in Section 1.5.

1.4.4.2.6 The functional organization of the cerebellum

Ridgway has addressed the large size of the cerebellum of the bottlenose dophin124. While not asproportionately large as in some species, the cerebellum of the dolphin is about 15% of the total brainmass compared to 10-11% in the human. Without a requirement for fine motor control of the digits ofthe hands, and less reliance on vision than in humans, the large size of the cerebellum suggestsexpanded powers with respect to the fine details related to hearing.

1.4.4.2.7 The functional organization of the cerebrum

Ridgway has also made a few remarks concerning the cerebral cortex or cerebrum of the dolphin. Little is known of its functional organization except a larger proportion of the auditory elements

Figure 1.4.4-5 The image map as projected onto the multi-dimensional correlator of the (visual) PGN. Whenprojected onto the foveola, the E fills only a small part of the complete correlator space. In fact, the correlator is ableto process both complete syllables simultaneously but also short words. The complete field in humans is about 150elements wide by at least 25 elements high. From Fulton, 2004.

5 September 2009101

125Ridgway, S. (1990) In Leatherwood, S. & Reeves, R. eds. Op. Cit. pg 76

appear to be located on the parietal lobe rather than across the sulcus in the parietal lobe (atopographic feature of little topological importance. Morgane, et. al. have also noted the less foldedcharacter of the cerebrum compared to the human. Again a topographic feature of little topologicalsignificance except to indicate a smaller overall area for the cerebrum within a given volume.

The primary topographic feature of the cerebrum is the substantially larger area devoted to hearingthan to vision.

1.4.4.3 The neural circuits of the dolphin neocortex

No major discussion of the organization of the dolphin neocortex, particularly as it participates inecholocation, could be found in the literature. Based on analyses by this author, the neocortexprobably plays a minor role in the acquisition, interpretation and perception of information acquiredby echolocation. Its role would be limited largely to the extraction of communications orientedinformation in the temporal lobes and cognitive functions associated with the frontal lobe of thehuman cortex.

The observation by Ridgway125 of the small number of commissure in the corpus callosum is probablyirrelevant in the absence of any discussion of the corpus principia within the diencephalon and theimportance of the lower brain stem in echolocation.

1.4.5 Overall features of the system

1.4.5.1 Potential acoustic imaging by dolphins

Conjecture has existed for many years over whether the dolphin can form a perceptual image of anacoustic target via its high frequency echolocation capability. Comments have also been documentedof some musical composers trying to describe the “image” of music they perceive during theircomposition activities. It is interesting to note that the human perceives full color images of hisexternal environment using totally abstract neural signals that do not reflect the actual wavelengthof the colors in the scene. The perception is developed using a visual correlator that appears virtuallyidentical to that used in human hearing. The perception is highly dependent on the early training ofthe subject. Thus, the human acoustic correlator could provide an “image” of music to one highlytrained in this field from a very young age. The diencephalon of the dolphin is significantly largerproportionately than that of humans. It appears entirely plausible that the dolphin, using a similarbut enhanced correlator, and following its early training, could perceive an image of a distant targetbased on its high frequency echolocation mechanisms. Similar conclusions can be drawn with regardto its lower frequency, wide beam echolocation capability (and even its entirely passive hearing).

1.4.5.2 The critical bandwidth in dolphins

The critical bandwidth characteristic provides significant support for the description of theinformation extraction architecture of the perigeniculate nucleus in hearing (and vision), whetherused in dolphins or other mammals. The most important aspect of the critical band is that it cannotbe described using a series of adjacent narrowband filters subdividing the auditory spectrum. Thecritical band always represents a bandwidth symmetrically placed about the center frequency of thetest frequency. If the test frequency is changed incrementally, the edges of the critical band alsomove incrementally.

5 September 2009102

126Lemonds, D. Au, W. Nachtigall, P. Roitblat, H. & Vlachos, S. (2000) High frequency auditory filtershapes in an Atlantic bottlenose dolphin J Acoust Soc Am vol 108, pg 2614(A)

The fundamental character of the critical band function has not been described previously in thehearing literature. That character is described in detail in Section 8.3.3 of “Processes in BiologicalHearing.” The critical band does not relate to a series of adjacent channels. It is always described asa bandwidth centered on the frequency of interest. As noted in Section 1.4.2.2, Lemonds, et. al.have provided critical bandwidth data for the bottlenose dolphin126. The critical bandwidth is ameasure of the correlation span of the associative correlator described in the previous section andshown at the top right in the above figure. Whereas the critical band in humans spans about 1/3octave (about 30% of the center frequency), it appears the dolphin only correlates over a span of about15% of the center frequency. This difference probably relates to the wider frequency range of thedolphin, in terms of the number of octaves covered.

1.4.5.3 The variation in time coherence within the system

The measurement of absolute and relative time delays plays a major role in the echolocation systemof dolphins. However, a variety of absolute and relative time delays are built into the auditorysystem itself. Understanding these time delays is critically important to the understanding of thesystems topology and physiology. The overall system involves a multitude of time delays. The majordelays when considering echolocation in dolphins are;

< associated with the finite velocity of sound in water, < associated with the finite velocity of acoustic energy within the cochlear partition,< associated with the finite velocity of neural signals traveling between processing engines, and< associated with the processing time required by an individual engine.

When considering the operation of the precision echolocation servo system (PES), it is also importantto consider the delays associated with the neurons and muscles associated with the sound generationsystem. It may also be important to understand the phase relationship between the signals producedby the two independent sources of ultra high frequency (UHF) sound available to the bottlenosedolphin.

In analyzing the performance of the auditory system of a given animal, it is often useful to develop atable of delays associated with every element in both the PES and the perceptual systems.

The neural portion of the auditory system of all mammals involves a large number of parallel neuralpaths. In many situations, the information is transferred in the bit-parallel, word-serial format. Inmany of these bit-parallel paths, the time delay associated with the individual neural paths are notthe same. At a given location and time, the signals on adjacent neurons associated with the samesignaling path may not apply to the same word. Such variations are inherent in the system. Theybegin with the slow propagation of the acoustic energy along Hensen’s stripe within the cochlea at anominal 6 m/sec (the nominal value for the human will be used in the absence of any other value). Asa result of this feature, the neural signals generated closest to the apex of the cochlea occur anominal 6 milliseconds behind the signals generated closest to the base. This variation is found inboth the tonal and the temporal channels of the initial neural system.

A similar skewing of the output data occurs in the visual system where it is used to advantage. Thedelay, relative to the earliest signal, of visual signals in the optic nerve is directly proportional to theangular distance of the target from the line of fixation of the eye. All of the signals in the optic nerve

5 September 2009103

127Zook, J. & DiCaprio, R. (1989) A potential system of delay-lines in the dolphin auditory brainstem InThomas, J. & Kastelein, R. eds. Sensory Abilities of Cetaceans. NY: Plenum Press pg 181-193

exhibit this skewness. It is due to the relative neural path length along the curved surface of theretina. The angular distance to each target in the visual field is extracted within the lateralgeniculate nucleus and used in both the alarm and awareness modes of perceptual operation and inthe pointing and convergence operations of the precision optical servomechanism. The skewness inthe signals is only removed, after the signals leave the lateral geniculate nucleus on their way to thecerebral cortex, by Meyer’s loop. Both the removal of the time dispersion and its introduction involvedistances of about 16 mm and signals propagating at the same velocity within the neural system.

In the binaural auditory system, the differential delay due to the difference in travel time of a signalfrom a target to the two ears must be considered. This differential delay is a function of both thedistance to the target and the angle to the target relative to the animals sagittal plane. In addition,the significant delay introduced by the cochlear partition must be considered. The slowness of theacoustic wave within the cochlear partition is apparent in the skewed output of both the temporalchannels and the tonal channels at the auditory nerve.

It is quite likely the auditory system (like the vision system) processes the neural signals data in bothskewed and de-skewed forms. However, complete de-skewing of the neural signals is much moredifficult in the auditory system than in vision. Using a simple delay line approach to de-skewingwould require differential line lengths of about seven times the length of the cochlear partitionbecause the acoustic velocity within the cochlear partition is at least seven times slower than in theneural system.

Zook & DiCaprio documented a de-skewing operation in the cochlear nucleus of the bottlenosedolphin127. They provide considerable commentary on the histology and cytology of the neurons in theventral cochlear nucleus (VCN) and the medial nucleus of the trapezoidal body (MNTB).Unfortunately, they did not place their results in a system context. It appears the variable length ofthe neurons observed between the VCN and the MNTB could in the same manner as Meyer’s loop invision. While Zook & DiCaprio did not measure the time delays associated with the observedneurons, they did provide estimates of the differential length of these neural paths. They report thisdistance varied between 3.4 to 4.5 mm in Stenella and 2.3 to 4 mm in Tursiops. They did not describethe neurons involved in detail. However, it can be assumed they were myelinated because of theiroverall length which Zook & DiCaprio estimated at 1.0 to 2.5 cm. Such neurons would be expected tobe myelinated, to include Nodes of Ranvier at intervals of about every millimeter or two, and operatein the phasic mode as part of the stage 3 signal propagation system. The speed of neuraltransmission in such neurons is typically 44 meters/sec. Thus, a length differential of 4 mm would beexpected to produce a time differential of about 0.1 msec. This is a small value relative to the totaldelay variation associated with the cochlear partition (nominal 36 mm divided by nominal 6 m/sec = 6msec). It would suggest the neurons observed were not associated with the auditory equivalent ofMeyer’s loop. Instead, they were probably performing a minor differential delay correction associatedwith the topography of the specific processing engines being interconnected.

Zook & DiCaprio did not present their calculations on time delay. However, they indicated calculateddelays about twice as large as presented here. Such values remain small relative to those associatedwith the cochlear partition.

5 September 2009104

128Morris, R. (1986) Op. Cit. Pg 388-391

129Bullock, T. & Gurevich, V. (1979) Soviet literature on the nervous system and psychobiology of cetaceaIntern Rev Neurobiol vol 21, pp 47-87

1.4.5.4 The auditory brain stem response

The auditory brain stem response (ABR) is an evoked potential (EP) very similar to that recordedfrom other mammals. Ridgway has shown the ABR recorded from the Dolphin, cat, monkey andhuman. These waveforms are analog waveforms. They can be decomposed into the individualwaveforms summed to form the ABR once the topology and physiology of the acoustic neural systemis understood (Fulton, 2005, Sec 11.1.6).

1.5 The echolocation performance of dolphins

It is difficult to assemble a composite estimate of the performance of dolphins from data collected in asingle time period. The data tend to come from different groups working at different times andlocations with instrumentation appropriate to that time period.

With such a sophisticated echolocation system, it is difficult to describe the overall performance ofthe system without relying on many caveats and footnotes. In general, the low frequencyperformance and the high frequency performance are totally independent. Furthermore, thereceiving system of the dolphin is difficult to model as a simple energy detector. While the inner haircells associated with the temporal channels are primarily envelope detectors, the outer hair cells areassociated with the tonal channels acting as a set of very narrow band energy detectors. The signalsfrom these channels are then correlated using multiple techniques to achieve different objectives. This is illustrated quite clearly by the critical bandwidth associated with the dolphins hearing. Thisnoise bandwidth is not associated with the overall bandwidth of the animal, nor is it the samefraction of the center frequency found in the human.

The maximum detection range for the dolphin echolocation system (under sea state 0), is generallyestimated at about 600 meters128. This figure can be assumed to apply to the low frequency, lowangular precision, auditory system using swept continuous tones. The angular precision would belimited to the binaural capabilities of the animal. The range precision would probably be limited tothe relative intensity of the reflection and the experience of the subject.

The range for echolocation using the high frequency regime is generally about 100 meters.

Bullock & Gurevich have reported on the performance of dolphins based on the Soviet literature upthrough the 1970's129. The values from Ayrapet’yants & Konstantinov (1970) are significant but notalways confirmed. They gave the data in Table 1.5.1-1 for the bottlenose dolphin.

5 September 2009105

130Au, W. (1989) Target detection in noise by echolocating dolphins In Thomas, J. & Kastelein, R. eds.Sensory Abilities of Cetaceans. NY: Plenum Press pp 203-216

131Floyd, R. (1986) Biosonar signal processing applications In Nachtigall, P. & Moore, P. eds. AnimalSonar: Processes and Performances. NY: Plenum Press pg 773-784

Table 1.5.1-1Echolocation Performance of the dolphin, T. truncatus

Discrimination of difference in distance = 1.5–4 mmThreshold for 75% correct response;simultaneous presentation; distance =1 meter or 3msec echo time.

Difference in echo time = 2-5 μsec

Minimum angular separation of targets:(Cylinders; 70% correct response)

Horizontal 14 arc minVertical 48 arc min

Horizontal (D. delphis) 1 arc min, 40 arc secVertical (D. delphis) 1 arc min, 40 arc sec

Au has presented data describing the range capability of the bottlenose dolphin using simplespherical metal targets and based primarily on its high precision pulse echolocation system130. Thedata has appeared in several of his subsequent writings.

Floyd has provided some performance data (primarily estimates) for the bottlenose echolocationsystem131. He has also compared that performance with two low performance man-made sonarsystems.

Morris has noted an interesting additional capability of the bottlenose dolphin (p. 389). At very closeranges (well within the near-field of the high frequency echolocation system) the subject is known toapproach man-made objects and encircle the object with its mouth without touching it. This appearsto be a method of precisely determining the diameter of an object.

Channel 5 of the British Television System has presented a documentary in its ExtraordinaryAnimals series (series 2, episode 4) named “The Dolphin Who Can See With His Ears.” The dolphinwas named Milo and has since died. It lived at the Aquarium in Belgium. Of course, the programdoesn’t recognize that all dolphins can image (see) with the echolocation portion of their auditorysystem.

- - - -

5 September 2009106

132Martin, S. et al & Houser, D. (2005) Instrumenting free-swimming dolphins echolocating in open waterJASA vol 117(4), pt 1, pp 2301-2307

133Houser, D. & Martin, S. et al. (2005) Echolocation characteristics of free-swimming bottlenose dolphinsduring object detection and identification JASA vol 117(4), pt 1, pp 2308-2317

Based on its anatomy and physiology, the dolphin exhibits a very significant interaural leveldiscrimination (ILD) capability over a nominal ±10 degrees in azimuth and ±5 degrees in elevation. Based on using only constant frequency pulses, the system achieves interaural time differences (ITD)small enough to discriminate between objects with range differences of about 1-2 cm. Nophysiological explanation for this capability has been found yet. If a swept frequency pulse was usedinstead, a relatively simple explanation of this range capability is available. The frequencydiscrimination capability of the system is not known with precision. However, experimental resultssuggest the dolphin can achieve spectral segregation levels on the order of 100 Hz in the 30-150 kHzrange. This range would encompass all of the Doppler velocities associated with its natural foodsources. By emitting a tone burst where the tone varies in frequency between 150 kHz and 30 kHz

and then returns to 150 kHz, the dolphin would be able to use this same spectral segregation level asa criteria in both range and velocity estimation calculations. The V-shaped tone burst would besimilar to the observed lower frequency whistles of the bottlenose dolphin. By comparing theinformation between the falling and the rising portion of the return, the Doppler velocity of the targetis effectively separated from the range related data. An estimate of the dolphins targeting capabilityis shown in Figure 1.5.1-2.

Martin et al. have developed new high performance instrumentation recently that allows a free-swimming dolphin to be tracked in al six degrees of freedom132,133. The first published results areilluminating, particularly comparing the different strategies used by dolphins in an obviouslyinadequate experiment planned by humans. The dolphins took shortcuts not anticipated by the

Figure 1.5.1-1 The targeting capability of the bottlenose dolphin, T. Truncatus ADD.

5 September 2009107

134Au, W. (2000) Echolocation in dolphins In Au, W. Popper, A. & Fay, R. eds. Op. Cit.pg 378

planner. Their spectrograms show the wide range of non harmonic frequencies employedsimultaneously during the hunt for a familiar instrumented target. The authors suggest the use oftwo independent click sound generators simultaneously. This author notes the simultaneousgeneration of multiple non-harmonic lower frequency sounds in the 20-40 kHz range (which arelabeled clicks in these papers but may not be based on spectrographs such as those of Boisseau indiscussed in Section 1.8.7.

1.5.1 Energy conservation in echolocation

The widely reproduced data of Au (1980) shows that the bottlenose dolphin employs a powerconservation strategy in its UHF signaling similar to that in modern cellular telephone systems134. The transmitter power is reduced following receipt of the first echo to a level that is optimal forinternal signal processing. The goal appears to be the same, minimize potential interference andreverberation while maintaining adequate signal-to-noise ratio.

1.6 The “fish stunning” performance of dolphins

Kassewitz has reported the first documented case of a dolphin intentionally stunning a fish(mullet)at a range of approximately two meters (private communications). They observed,videographed and recorded the acoustic actions of the animal. The acoustic signature of the attack isreproduced in Figure 1.6.1-1.

5 September 2009108

The frequency spectrum of this 40 kHz waveform is somewhat unexpected since the beam formingmechanism available at frequencies in the 120-150 kHz region are not believed to be available in thislower frequency range. If true, the power density at a distance of two meters forward of the animalwould be expected to be considerably lower than that achievable at higher frequencies for the sameacoustic generator power. The two meter range is just beyond the estimated near-field radiationpattern of the 150 kHz beam forming mechanism. The nominal beam-forming gain for a ten-degreesymmetrical beam is approximately 26 dB (412:1) over the isotropic condition.

- - - - -END

8.1.6.2 Potential range determination based on frequency dispersion

By employing surface acoustic wave technology applied to a liquid crystalline surface within thecochlea, frequency dispersion is avoided within the hearing system. However, frequency dispersiondoes occur in any Newtonian fluid forming the exterior medium, whether gaseous or liquid.

Frequency dispersion in the exterior environment can be used in hearing to estimate the range to asource. The effect is somewhat like the Doppler Effect. In the Doppler Effect, the pitch of a signal

Figure 1.6.1-1 Record of an acoustic stunning attack by a bottlenose dolphin. Two twelve millisecond bursts of midfrequency energy centered on 40 kHz. Upper frame shows relative power in MU versus time. Lower frame showsthe energy quite widely distributed in frequency with a potential harmonic relationship with a fundamental of about10 kHz. From Kassewitz, 2007.

5 September 2009109

135Lighthill, J. (1978) Waves in Fluids. NY: Cambridge Univ Press pp 235-245

changes as a function of the velocity of the source. In this case, the individual frequency componentsof an emitted sound travel through the medium at different phase velocities. This effect causes theenvelope of the emitted sound to change with distance. It also causes the higher frequencycomponents to reach a given point before the lower frequency components. If the auditory system isdesigned to detect the time of arrival of the individual frequency components of a well characterizedsignal, the system can estimate the range of the source. Lighthill develops the equations relevant tothe dispersion of a bulk acoustic wave in a fluid but does not give practical values for the dispersionof sea water135.

A potential exists for a whale to estimate the range to another whale emitting a standardized callthrough the frequency dispersion of the seawater. A brief literature search did not uncoverdispersion data for very low frequency (well below 1000 Hz) acoustic dispersion in seawater.

5 September 2009110

Table of Contents September 5, 2009

.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.0 A case study in echolocation and communications in the dolphin . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Overview of hearing and active echolocation in Dolphins . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.1 Species specific background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.1.1.1 Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.1.1.2 Brief Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.1.1.3 Technical environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.1.1.3.1 The acoustic environment of sea mammals . . . . . . . . . . . . 111.1.1.3.2 The frequency range of dolphin emissions . . . . . . . . . . . . . 131.1.1.3.3 The dispersion of water-based liquids . . . . . . . . . . . . . . . . 131.1.1.3.4 Range and Range resolution in sea water . . . . . . . . . . . . . 141.1.1.3.5 Doppler frequency versus velocity in sea water . . . . . . . . . 15

1.1.1.4 Echolocation in Humans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151.1.2 Overall vocal and auditory physiology of the dolphins . . . . . . . . . . . . . . . . . . . 15

1.1.2.1 Modification of the non-neural auditory system . . . . . . . . . . . . . . . . 181.1.2.1.1 Introduction to the outer ear of the dolphin . . . . . . . . . . . . 181.1.2.1.2 Introduction to the outer-to-middle (or inner) ear interface in

the dolphin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191.1.2.1.3 Introduction to the inner ear in the dolphin . . . . . . . . . . . 20

1.1.2.2 The expansion of the auditory neural system . . . . . . . . . . . . . . . . . . 211.1.2.3 The expansion and optimization of the phonation system . . . . . . . . 23

1.1.2.3.1 The composition of the melon in the dolphin . . . . . . . . . . . 231.1.2.3.2 The efficient generation of sound by Cetacea . . . . . . . . . . . 231.1.2.3.3 Alternate methods of high frequency sound generation . . 241.1.2.3.4 The generation of the high frequency sound fields . . . . . . 241.1.2.3.5 The potential generation of dual click sound fields . . . . . . 241.1.2.3.6 An alternate schematic for the generation of the high

frequency sound fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251.1.2.3.7 Acoustic propagation within the melon and other cavities of

the dolphin head . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291.1.2.3.8 Pulse repetition rates in high frequency echolocation . . . . 29

1.1.3 Overview of signaling in dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311.1.3.1 The external anatomy and beam forming capability of the outer ears

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321.1.3.2 The external anatomy and beam forming capability of high frequency

vocalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391.1.3.3 Typical signaling waveforms of the dolphin . . . . . . . . . . . . . . . . . . . . 39

1.1.3.3.1 Click-type waveforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391.1.3.3.2 Whistle-type waveforms . . . . . . . . . . . . . . . . . . . . . . . . . . . 411.1.3.3.3 Truncatus waveforms recorded in the Sado estuary of Spain

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441.1.3.3.4 Truncatus waveforms & spectra from the Irish Sea &

Shannon estuary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471.1.3.3.5 Truncatus waveforms from Djurpark, Sweden . . . . . . . . . 49

5 September 2009111

1.1.3.3.6 Truncatus waveforms from Sarasota Bay, Florida . . . . . . 491.1.3.3.7 Characteristics of tonal vs Stressed languages . . . . . . . . . 491.1.3.3.8 Mimicry by dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501.1.3.3.9 The concept of an individual dolphin adopting a personal

noun, limited to a signature whistle . . . . . . . . . . . . . . . . . . . 511.2 The detailed description of the echolocation and communications system of the dolphin

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521.2.1 The acoustic energy generating capability of the dolphin . . . . . . . . . . . . . . . . . 52

1.2.1.1 Low frequency sound generation in dolphins RESERVED . . . . . . . . 531.2.1.2 Ultra high frequency sound generation in dolphins . . . . . . . . . . . . . 531.2.1.3 High frequency sound generation in dolphins . . . . . . . . . . . . . . . . . . 54

1.2.2 The beam forming capability of the dolphin . . . . . . . . . . . . . . . . . . . . . . . . . . . 541.2.2.1 High frequency, wide beam, echolocation in dolphins . . . . . . . . . . . . 55

1.2.2.1.1 The anatomy of HF echolocation in dolphins RESERVED. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

1.2.2.1.2 The schematic of HF echolocation in dolphins RESERVED. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

1.2.2.2 Ultra high frequency, narrow beam, echolocation in dolphins . . . . . 551.2.2.2.1 The anatomy of UHF echolocation in dolphins . . . . . . . . . 551.2.2.2.2 The potential for rapid beam steering . . . . . . . . . . . . . . . . 59

1.2.2.3 Potential operation in the 150-390 kHz . . . . . . . . . . . . . . . . . . . . . . . 591.2.3 The range measuring capability of the dolphin . . . . . . . . . . . . . . . . . . . . . . . . . 60

1.2.3.x The related range measuring capability of the bats . . . . . . . . . . . . . 601.3 The overall neural physiology of echolocation in dolphins . . . . . . . . . . . . . . . . . . . . . . . . 63

1.3.1 Intrinsic signal generating capability of the dolphin . . . . . . . . . . . . . . . . . . . . 651.3.2 Potential signals received by the dolphin from its environment . . . . . . . . . . . 66

1.3.2.1 Character of exogenous signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671.3.2.2 Character of the reflected endogenous signals . . . . . . . . . . . . . . . . . 67

1.3.2.2.1 The reflective characteristics of fish . . . . . . . . . . . . . . . . . . 671.3.2.2.2 The character of reflected continuous tones . . . . . . . . . . . . 681.3.2.2.3 The character of reflected impulse signals . . . . . . . . . . . . . 711.3.2.2.4 Useful biosonar configurations . . . . . . . . . . . . . . . . . . . . . . 71

1.3.3 Extensions based on binaural and biphonic implementations . . . . . . . . . . . . . 711.3.4 The potential neural architecture of the dolphin . . . . . . . . . . . . . . . . . . . . . . . . 72

1.4 The neural physiology of specific circuits used for echolocation in dolphins . . . . . . . . . . 731.4.1 Overview of the electrolytic neuron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

1.4.1.1 Basic physiological forms of neurons . . . . . . . . . . . . . . . . . . . . . . . . . 731.4.1.2 The bandwidth and propagation velocity of myelinated neurons . . . 74

1.4.2 The acoustic receiving & signal generation system in dolphins . . . . . . . . . . . . 741.4.2.1 Technical background & definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 741.4.2.2 Details of acoustic energy reception in the dolphin . . . . . . . . . . . . . . 75

1.4.2.2.1 A physiologically compatible lens antenna for reception . 751.4.2.2.2 The potential for the teeth to form an antenna in dolphins

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 771.4.2.3 Details of the frequency selection mechanism in the dolphin . . . . . . 771.4.2.4 Details of neural signal generation in the dolphin . . . . . . . . . . . . . . 80

1.4.2.4.1 The dispersion of the received energy within the cochlea of dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

1.4.2.4.2 Initial signal generation by phonoreceptors in the dolphin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

5 September 2009112

1.4.2.4.3 Energy vs phase sensitivity of the signal generationmechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

1.4.2.5 Composite performance of the dolphin . . . . . . . . . . . . . . . . . . . . . . . . 841.4.2.5.1 The limiting auditory sensitivity of dolphins . . . . . . . . . . . 841.4.2.5.2 The frequency discrimination capability of the dolphin . . 85

1.4.3 Signal processing & signal propagation in dolphins . . . . . . . . . . . . . . . . . . . . . 871.4.4 Signal manipulation and cognition in dolphins . . . . . . . . . . . . . . . . . . . . . . . . . 88

1.4.4.1 The gross morphology, homeostasis and vascular circulation . . . . . 881.4.4.1.1 The internal architecture of the engines of the dolphin brain

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911.4.4.2 The neural circuits of the dolphin paleocortex EDIT . . . . . . . . . . . . 92

1.4.4.2.1 The functional organization of the cochlear nucleus . . . . . 921.4.4.2.2 The functional organization of the lateral lemniscus

(Reserved) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 921.4.4.2.3 The functional organization of the inferior colliculus . . . . 921.4.4.2.4 A conceptual inferior colliculus in the dolphin . . . . . . . . . 961.4.4.2.5 The functional organization of the perigeniculate nucleus

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 971.4.4.2.6 The functional organization of the cerebellum . . . . . . . . 1011.4.4.2.7 The functional organization of the cerebrum . . . . . . . . . . 101

1.4.4.3 The neural circuits of the dolphin neocortex . . . . . . . . . . . . . . . . . . 1021.4.5 Overall features of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

1.4.5.1 Potential acoustic imaging by dolphins . . . . . . . . . . . . . . . . . . . . . . 1021.4.5.2 The critical bandwidth in dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . 1021.4.5.3 The variation in time coherence within the system . . . . . . . . . . . . 1031.4.5.4 The auditory brain stem response . . . . . . . . . . . . . . . . . . . . . . . . . . 105

1.5 The echolocation performance of dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1051.5.1 Energy conservation in echolocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

1.6 The “fish stunning” performance of dolphins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

- - - - -END . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1098.1.6.2 Potential range determination based on frequency dispersion . . . . 109

5 September 2009113

Appendiox L List of Figures 9/5/09

Figure 1.1.1-1 High frequency sound attenuation in sea water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Figure 1.1.1-2 Ambient noise in the ocean measured in 1/3 octave bands . . . . . . . . . . . . . . . . . . . . 13Figure 1.1.1-3 Range equation and range resolution in sea water . . . . . . . . . . . . . . . . . . . . . . . . . . 14Figure 1.1.1-4 Doppler frequency shift as a function of relative velocity in sea water . . . . . . . . . . . 15Figure 1.1.2-1 Top level block diagram of vocal and auditory systems in dolphin . . . . . . . . . . . . . . 16Figure 1.1.2-2 Three discrete lobes of highly differentiated fats have been identified . . . . . . . . . . . 19Figure 1.1.2-3 The outer/inner ear interface of the dolphin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Figure 1.1.2-4 The generic mammalian physiology of the auditory system applicable to the dolphin

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22Figure 1.1.2-5 The dual channel operation of the sound generation system in the bottlenose dolphin

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27Figure 1.1.3-1 The outer ear of the bottlenose dolphin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Figure 1.1.3-2 A front view of the bottlenose dolphin showing the forward looking lens-type external

ears . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 1.1.3-3 Estimated field patterns of the auditory receivers of the bottlenose dolphin at 120

kHz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Figure 1.1.3-4 Temporal and spectral characteristics of a high frequency click . . . . . . . . . . . . . . . . 39Figure 1.1.3-5 A nearly optimal click with a fundamental frequency of 78 kHz . . . . . . . . . . . . . . . . 41Figure 1.1.3-6 A nominal high frequency (HF) signal spectrogram generated by dolphin . . . . . . . . 42Figure 1.1.3-7 A collage of LF spectrograms from two Atlantic bottlenose dolphins . . . . . . . . . . . . 44Figure 1.1.3-8 Characteristics and occurrences of whistles in Sado estuary . . . . . . . . . . . . . . . . . . 46Figure 1.1.3-9 T. Truncatus waveforms identified by Hickey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Figure 1.2.2-1 The echolocation geometry of the bottle-nosed dolphin . . . . . . . . . . . . . . . . . . . . . . . 56Figure 1.2.2-2 The general layout of the echolocation system of the dolphin . . . . . . . . . . . . . . . . . . 58Figure 1.2.2-3 Potential beam direction variation in the bottlenose dolphin . . . . . . . . . . . . . . . . . . 59Figure 1.2.3-1 An example of a fruit fly pursuit by C. Psilotis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Figure 1.3.1-1 A beluga whale click interval pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Figure 1.3.1-2 Waterfall displays of the energy distribution in emitted clicks with a 205 dB threshold

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Figure 1.3.2-1 The characteristics of various reflected signals expected in biosonar . . . . . . . . . . . . 70Figure 1.4.2-1 The outer ear and field patterns of the bottlenose dolphin . . . . . . . . . . . . . . . . . . . . 76Figure 1.4.2-2 A comparison of dolphin and human cochlear curvatures . . . . . . . . . . . . . . . . . . . . 78Figure 1.4.2-3 The “cranked” form of the cochlear partition in Delphinidia . . . . . . . . . . . . . . . . . . . 79Figure 1.4.2-4 The acoustic information extraction mechanism of hearing . . . . . . . . . . . . . . . . . . . 81Figure 1.4.2-5 A typical dolphin cochlear partition overlaid by a Hankel function . . . . . . . . . . . . . 82Figure 1.4.2-6 Performance parameters for Hensen’s stripe for the nominal dolphin cochlea . . . . 83Figure 1.4.2-7 Sensitivity of dolphin compared to human hearing . . . . . . . . . . . . . . . . . . . . . . . . . . 85Figure 1.4.2-8 Frequency difference threshold for the bottlenose dolphin and humans . . . . . . . . . 86Figure 1.4.4-1 Volumes of auditory and visual structures in the brain of dolphins . . . . . . . . . . . . . 91Figure 1.4.4-2 The characteristics of various reflected signals expected in biosonar . . . . . . . . . . . . 95Figure 1.4.4-3 Potential organization of the auditory analysis function in dolphins . . . . . . . . . . . . 98Figure 1.4.4-4 The human Spiral Tonal Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100Figure 1.4.4-5 The image map as projected onto the multi-dimensional correlator of the (visual) PGN101Figure 1.5.1-1 The targeting capability of the bottlenose dolphin, T. Truncatus ADD . . . . . . . . . 107Figure 1.6.1-1 Record of an acoustic stunning attack by a bottlenose dolphin . . . . . . . . . . . . . . . . 109

5 September 2009114

SUBJECT INDEX (using advanced indexing option)

ABR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88, 105acoustic environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11, 21action potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79activa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74adaptation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 74auditory nerve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 79, 104axon segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74broadband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26, 38, 39, 53caecum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18cAMP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2cerebellum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101cerebrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101, 102chord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68cochlear nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 9, 87, 91, 92, 104cochleotopic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63colliculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 30, 60, 72-74, 79, 84, 90, 92, 93, 96, 97computational anatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98curvature of Hensen’s stripe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 75, 80-82diencephalon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72, 73, 79, 88, 89, 102diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74dynamic range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66echolocation . . 1-3, 5, 8, 11, 13-17, 21, 23, 25, 26, 28, 34, 39, 42, 45-47, 50, 52, 54-60, 62-64, 66, 67, 71-

74, 77, 86, 87, 89-92, 97, 98, 102, 103, 105-108evoked potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105evoked potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84Gaussian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 57Hankel function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78, 82, 93, 96Hensen’s stripe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 75, 77, 78, 80-83, 103homologs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72inferior colliculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 30, 60, 72-74, 79, 84, 90, 92, 93, 96, 97inhomogeneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11interp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97, 98lateral geniculate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97, 104launcher . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81medial geniculate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 79, 97mesencephalon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93, 97monopulse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2, 54, 71MRI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2, 6, 7, 9multi-dimensional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93, 96, 97, 101myelinated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74, 104Myelination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74narrow band . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39, 41, 66, 105neurite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74neurites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74, 98Newtonian fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14, 109

5 September 2009115

Node of Ranvier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 8, 11, 13, 14, 31, 38, 52-54, 57, 64, 79, 97, 98, 105, 106, 108operculum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21parietal lobe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102percept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98perigeniculate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72, 79, 97-99, 102perigeniculate nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72, 97-99, 102perilymph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81PES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16, 73, 74, 92, 103phonoreceptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84, 87phylogenic tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89place-frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92place-frequency-delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92poditic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74pons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91POS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87, 92precision optical servomechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104propagation velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74, 87pulse-to-pulse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60pulvinar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72, 91, 98resonance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9, 25, 28, 35, 53, 54, 73segregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93, 96, 107servo loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4, 16servomechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4, 60, 61, 104signal-to-noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57, 98, 108signal-to-noise ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98, 108spinal cord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9spiral ganglia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 30, 38, 74, 87, 88spiral tonal map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99, 100stage 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79, 93, 96, 104stage 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16synapse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74tectorial membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 21, 57, 75, 78, 81, 84temporal lobe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72thalamic reticular nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93, 97thalamus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72, 73, 91timbre . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52tonotopic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63topography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97, 103, 105type I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78type II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78vestibule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20, 21, 57, 76, 81, 82